ISTA fluxonium design switches bit-flip protection on demand with 10 ns gates

[Neuer]

Researchers at the Institute of Science and Technology Austria have proposed a clever new design for fluxonium superconducting qubits. Quantum Zeitgeist covered the paper by C. A. Siegele, A. A. Sokolova, L. N. Kapoor, F. Hassani and J. M. Fink. The key idea is a qubit whose protection against bit flip errors can be turned up or down in real time

That solves a long standing dilemma. Heavily protected qubits are great at storing information, but they are hard to control quickly, because the same protection makes them resistant to the signals used to manipulate them. Lightly protected qubits are easy to control but lose information faster. This design lets one qubit switch between the two modes

In the heavy setting, the qubit's energy relaxation time could in principle reach hours. In the light setting, it can be controlled quickly. The difference in relaxation times between the two modes is roughly 100,000 to 1,000,000 times. The switch works by using magnetic flux to tune either the Josephson energy or the charging energy, or in a second version by switching in a large capacitance

The gates are remarkably fast. Single and two qubit gates use flux pulses of about 10 nanoseconds, a speed that usually requires microwave control. A simulated single qubit gate reached an infidelity of 2 x 10^-5 with a 7.5 nanosecond pulse. The team expects dephasing, rather than relaxation, to be the main limit

These are simulation results rather than a hardware demonstration, so the next step is building the device. If it works as predicted, protection on demand could be a very useful tool for superconducting quantum computers, combining long memory with fast processing in the same qubit

Could switchable protection become standard in superconducting qubit design? And how much do you trust simulated gate fidelities before hardware results?

Superconducting Qubits Now Switch States In Just 10 Nanosecondsquantumzeitgeist.com

[William]

Ten nanosecond gates driven by flux alone is fast. Removing microwave control lines would simplify the wiring inside the fridge. That alone could help scaling. Fewer cables also means less heat leaking into the fridge

[MrRicardo]

Relaxation times of hours would be extraordinary. Today's best superconducting qubits manage around a millisecond

Even a fraction of that improvement would be a big deal. Dephasing is the catch, as the authors admit

Memory qubits with that kind of lifetime would change how algorithms are designed. Long idle periods would stop being a problem. That opens up a lot of possibilities

[SoloOrca]

Johannes Fink's group at ISTA has a strong record in superconducting circuits. When they say something should work, it usually does. Looking forward to the experiment

Austria keeps producing excellent quantum physics. ISTA has become a real powerhouse

[Quiet Forge]

Protection on demand is such an elegant idea. Store information in a heavily protected state, then loosen it when you need to compute. It is like locking a safe and only opening it when you need something

Classical computers do something similar with memory and registers. Fast but fragile for working, slow but safe for storing

[Jude_54]

Simulated infidelity of 2 x 10^-5 is impressive, but hardware always adds noise you did not model. Flux noise in particular can be nasty. I would expect the real number to be a bit worse

Still a great target. Even ten times worse would still be excellent

[NovaPrime90]

Is switching between modes itself a source of errors? Every operation adds risk

If the switch is clean, this works brilliantly. If not, the benefits could shrink. Leakage out of the computational space during the switch is the obvious risk. Careful pulse shaping might handle it