Picosecond pulses offer a faster control route for fluxonium qubits

The control problem

Quantum computers use precisely timed operations, called gates, to steer their qubits. Fluxonium is a superconducting qubit design with a strongly nonlinear energy spectrum. That structure can help limit unwanted transitions out of the two states used for computation. But fluxonium’s low transition frequency creates a control challenge: pulse sequences developed for the more common transmon qubit may be inefficient for it.

The paper studies single-flux-quantum (SFQ) pulses: very brief electrical pulses generated by Josephson-junction circuits. Instead of shaping each control signal as a conventional microwave waveform, SFQ control uses digital circuits to deliver timed pulses. The researchers ask whether arranging these pulses differently can make control both fast and accurate for fluxonium.

Two ways to arrange the pulses

The authors first examine regular SFQ pulse trains sent through either a flux-control channel or a charge-control channel. In their simulations, these straightforward sequences involve a trade-off: flux-only control is relatively slow, while charge-only control can excite higher energy levels. Such unwanted transitions out of the computational states are known as leakage.

They then simulate two alternatives using a model that includes ten fluxonium energy levels. The first packs pulses at a high frequency and repeats each packet once per qubit period, sending it through the flux channel. The timing is designed to reinforce the desired transition while reducing excitation of some unwanted transitions. For one representative simulated qubit, the authors report an Xπ gate—a rotation that swaps the two computational states—with 99.995% fidelity, leakage below 0.01%, and a duration of 68 nanoseconds.

The second approach sends pulse packets through both flux and charge channels. The sequences are offset by a quarter of the qubit period, and the numbers of pulses in the two channels are optimized separately. In the model, this arrangement strengthens the resonant drive and reduces a source of control error called the counter-rotating contribution. Across three representative sets of qubit parameters, simulated Xπ gates take about 17–27 nanoseconds. The paper also reports Xπ/2 and Yπ/2 rotations—quarter-turn, or half-π, rotations—lasting about 8.5–13.5 nanoseconds. Calculated fidelities for these examples are at least around 99.99%.

These are simulation results, not measurements from a working fluxonium device controlled by the proposed protocols. The calculations describe how the modeled qubit responds to specified pulse sequences; they do not show that a physical system will achieve the same fidelities under real operating conditions.

What remains uncertain

The authors test the flux-only protocol against moderate changes in modeled qubit parameters and flux bias. They report that it remains within their chosen fidelity criterion over the variations examined. The combined protocol is evaluated for three parameter sets. These tests indicate robustness within the simulations, not a guarantee across devices or operating conditions.

Several practical effects are left for future work, including timing differences between the two control channels, pulse distortions, and dissipative processes. The authors also identify experimental benchmarking and extensions to two-qubit gates as next steps. Although the paper discusses possible circuitry, it does not demonstrate a complete hardware implementation of either protocol.

This is an arXiv preprint, and its claims should be treated as preliminary rather than as settled results from peer-reviewed publication. The reported fidelity figures are outcomes of the authors’ simulations and assumptions—not measured performance or proof that SFQ control is ready for scalable quantum processors.