One Pulse Now Builds a Quantum Code, Not Thousands

Inside a superconducting circuit chilled near absolute zero, a fragile quantum state has only a sliver of time before a stray vibration, a spare photon or a whisper of electrical noise scrambles the information it carries. Physicists at Chalmers University of Technology in Sweden and Tianjin University in China have found a way to build these error-resistant quantum states, called bosonic codes, in a single pulse rather than the thousands of pulses earlier methods required. The trick, described in the journal Physical Review Letters, is mathematical rather than mechanical: instead of nudging a state toward its target step by step, the team calculates the one tailored pulse that gets there directly. That shortcut cuts the window for disaster by more than 1,000 times, at least on paper and in simulation.

Quantum computers promise breakthroughs in drug design, cryptography and logistics that ordinary computers cannot touch, but only once they stop losing their nerve at the first sign of trouble. Every qubit, whether built from a trapped ion, a photon or a loop of superconducting wire, is exquisitely sensitive to the outside world, and every extra moment spent processing information is another moment for an error to sneak in, a dynamic Google’s own experiments scaling up error-correcting superconducting qubits have made explicit.

One of the more promising defenses against those errors does not store information in a single qubit at all. Instead, so-called bosonic codes spread a bit of quantum information across the many possible energy states of a microwave field inside a superconducting cavity, the same kind of circuit at the heart of Chalmers’ own quantum hardware. Spreading the information out this way builds in redundancy: knock the system slightly off course, and there is enough structure left over to notice the slip and correct it, much as a well-designed error-correcting code catches a flipped bit on a hard drive. The catch is that building one of these delicate states from scratch, and then manipulating it, has always been a slow, deliberate process.

Earlier methods relied on adiabatic ramping, a technique borrowed from decades of quantum-control research that nudges a system toward its target so gently that it barely notices the change. Gentleness has a cost: those ramps typically stretch across thousands of driving cycles, and every one of those cycles is another chance for a stray photon or a flicker of electrical noise to spoil the result.

A Different Kind of Shortcut

Tangyou Huang, Lei Du and their co-author Lingzhen Guo of Tianjin University realized the slow ramp was never mathematically necessary. Their method calculates the exact driving pulse, a single carefully shaped burst of microwave energy, needed to carry a quantum system from its starting point directly to the target state within one Floquet period, the natural repeating cycle of a periodically driven system. In simulation, that single-period approach built the field’s standard test states, including the four-lobed cat state and other bosonic codes, directly from an empty cavity, with fewer than 1 error in every 100,000 attempts, a result that sits alongside other periodic-driving approaches independent groups have proposed for related error-resistant states.

“You can think of it like building a large Lego castle,” says Tangyou Huang, researcher in Quantum Technology at Chalmers and co-author of the study. “Instead of assembling it brick by brick and risking mistakes along the way, quantum lattice gates act like pre-built Lego modules that can be connected quickly and efficiently.”

The building blocks behind the shortcut are what the team calls quantum lattice gates, a toolkit the same three researchers proposed previously, which leans on the full nonlinearity of the Josephson junctions already built into superconducting circuits, rather than the simplified slice of that nonlinearity such circuits usually rely on. Instead of turning knobs to gradually reshape a wave over time, the team works backward from the desired final state, calculates the driving pulse that produces it mathematically, and slices that pulse into a rapid sequence of lattice-gate operations that reproduce it almost exactly within a single cycle. To check whether the approach faithfully reproduces the full range of quantum behavior, rather than some narrow slice of it, the researchers benchmarked it against so-called Haar-random states, the standard yardstick physicists use to confirm that an operation behaves as unpredictably, and therefore as universally, as true randomness demands. The simulated results tracked the expected statistics closely across a range of system sizes, evidence the shortcut is not cutting corners on the states it can reach. The team also ran basic logical operations, the quantum equivalent of flipping or rotating a bit, on top of these codes with roughly 1 error in every 10,000 operations, a rate the authors say is comparable to some of the best superconducting qubits built in the lab today.

Still Waiting for Real Hardware

None of this has yet run on real hardware. Everything reported so far comes from numerical simulation of an idealized cavity, and translating a mathematically exact pulse into an imperfect microwave generator, one with its own noise and drift, is its own engineering problem.

The team did model how the approach copes with two common imperfections, random fluctuations in pulse strength and in driving frequency, and found the optimized version keeps its fidelity above 99% for total disturbance levels below roughly one part in ten, a margin that comfortably outlasts the far slower adiabatic method it replaces. Still, real superconducting circuits carry other quirks, from fabrication variation to thermal drift, that a simulation cannot fully anticipate.

If the pulse survives contact with real hardware, the payoff is not just speed for its own sake. Every driving cycle removed from a quantum operation is a driving cycle’s worth of exposure to decoherence removed with it, which is precisely the accounting that determines whether a fault-tolerant quantum computer is buildable at all. The method is also, by design, compatible with the same superconducting circuit platforms already in use, including the 100-qubit machine Chalmers has under development, rather than demanding some entirely new piece of hardware. Huang and Du say they are already discussing an experimental demonstration with colleagues at Chalmers.

A single pulse, precisely aimed, is a very different proposition from a thousand small nudges in the dark, and the difference may be exactly what separates a quantum computer that occasionally works from one that can be trusted to run for as long as a real calculation takes. For now, that trust still has to survive its first trip off the simulation and onto an actual chip.

FAQ