A Quantum Operation That Took Thousands of Driving Cycles Now Takes One. Chalmers Says That Is a 1,000-Fold Speedup for Error-Protected Qubits.

One of the more promising routes to a quantum computer that does not drown in its own errors is to stop storing information in individual qubits at all. Instead, so-called bosonic codes encode a logical qubit in the state of a microwave field inside a superconducting resonator, spreading the information across many photons so that the most common physical errors, such as losing a single photon, can be detected and undone. The catch has been speed: building and manipulating those field states has required driving the system through thousands of repeated control cycles, each one an opening for noise. Researchers at Chalmers University of Technology in Sweden now report a way to do it in one.

"Our method shows that a diverse range of quantum operations on bosonic states can be completed within a single driving cycle, rather than the several thousand cycles that have been required previously," said Lei Du, a researcher in applied quantum physics at Chalmers and lead author of the paper, published September 10 in Physical Review Letters under the title "Single-Period Floquet Control of Bosonic Codes with Quantum Lattice Gates." The improvement is a factor of more than 1,000 in the time an operation occupies.

The approach rests on what the team calls quantum lattice gates, a universal set of operations tailored to the periodic, or Floquet, driving that superconducting circuits already use. "Previous approaches that build up such quantum states one piece at a time have required quantum systems to be guided through thousands of repeated driving cycles, a time-consuming process that leaves more opportunity for errors to occur," Du said. Co-author Tangyou Huang, a researcher in quantum technology at Chalmers, offered an analogy. "You can think of it like building a large Lego castle," he said. "Instead of assembling it brick by brick and risking mistakes along the way, quantum lattice gates act like prebuilt Lego modules that can be connected quickly and efficiently."

The practical payoff is in the error budget. Every quantum operation competes against decoherence, the steady leakage of quantum information into the environment, and an operation that finishes 1,000 times sooner gives the environment 1,000 times less opportunity to intervene. "This makes the operations both faster and more efficient, while reducing the risk that disturbances will corrupt the information before the process is finished," Du said. "It represents an important step toward fault-tolerant quantum computers." Bosonic codes have already been shown to protect against certain error types more effectively than arrays of ordinary qubits, Huang noted, which is why groups at Yale, Amazon Web Services and elsewhere have pursued them; the Chalmers result addresses the speed penalty that has held them back.

The work is theoretical, but it was designed with a specific machine in mind. Chalmers is building a 100-qubit superconducting quantum computer, and the microwave resonators and drive electronics the scheme requires are standard parts of that platform. "A key advantage of our approach is that it can be implemented using existing superconducting quantum circuit platforms," Huang said. "We are already discussing possible experimental realizations with colleagues at Chalmers, and we hope to see a demonstration of the method in the near future."

If the demonstration works, it would remove one of the practical objections to bosonic encoding and strengthen the case that the path to useful quantum computing runs through fewer, better-protected logical qubits rather than ever-larger arrays of fragile physical ones. The applications the team cites, drug discovery, energy systems, cryptography and logistics, are the usual ones. The difference here is a concrete reduction, by three orders of magnitude, in how long the hardware has to hold its breath.