Physicists propose reading a qubit through the tiny supercurrent it makes itself, reaching 99.85% accuracy in simulations

A qubit built from a quantum dot between two superconductors could be read by watching the tiny oscillating supercurrent it produces on its own, with no spare qubit drafted in to take the measurement. In numerical simulations of that scheme the reconstructed state matched the true one with 99.52%, 99.81% and 99.85% accuracy, depending on how many current samples were taken. All of it is theory so far, and no one has built the circuit.

The day in tech, in one email: TLDR is the daily newsletter with links and TLDRs of the most interesting stories in startups, tech and programming.

Get TLDR →

We may earn a commission if you sign up.

The idea comes from physicists in Brazil and China, led on the Brazilian side by José Carlos Egues of the São Carlos Institute of Physics at the University of São Paulo and on the Chinese side by Yugui Yao of the Beijing Institute of Technology. Their goal is a readout that disturbs the qubit as little as quantum mechanics permits, because reading a quantum bit normally collapses the very superposition that makes it useful, and every extra qubit or reset step spent on the readout is hardware and time that a larger processor would have to pay for many times over.

Reading an Andreev qubit from its own interference current

The qubit in question is an Andreev qubit, which lives in a quantum dot sitting inside an indium arsenide nanowire that links two superconducting leads. According to the ScienceDaily write-up of the FAPESP release, a qubit in superposition drives a supercurrent with an oscillating component. The steady part of that current and the size of the oscillation encode how much weight each basis state carries, while the moment the oscillation starts encodes the phase between them.

The authors call the effect an AC-like Josephson effect without any applied voltage, a phrase from the abstract of their arXiv preprint, “Andreev qubit readout from dynamic interference supercurrent”. The oscillation arises from quantum interference between the junction’s many-body energy states, so the qubit is effectively broadcasting its own condition. A SQUID, a very sensitive magnetic-flux sensor, sits near the circuit and picks up the field from that circulating current.

Egues told Agência FAPESP that the aim is to pull the information from a signal produced by the qubit under study itself, in place of shipping it to an auxiliary qubit first. Conventional nondemolition schemes lean on those ancillas and on repeated reset cycles to prepare them again. The new proposal says both can be skipped, and that measurement can continue while the qubit keeps evolving, Agência FAPESP’s account of the work explains.

Simulated accuracy at three, five and nine samples per cycle

The three accuracy figures come from simulations that folded in measurement noise and the small back-action of the detection coil on the qubit. A separate review at PostQuantum, which read the Letter closely, translates them into readout infidelities: 0.48% for three measurements per oscillation period, 0.15% for five and 0.19% for nine. Subtracted from 100%, those give 99.52%, 99.85% and 99.81%, so the best result, 99.85%, came with five samples rather than the most.

The authors say the numbers sit within thresholds used in quantum error correction. PostQuantum disputes how meaningful that comparison is, arguing that the infidelity of a fitted state is a different quantity from the per-operation error rates those thresholds describe. The same review also questions whether the assumed sampling interval and energy gap fit the 21.5 GHz qubit frequency in the paper, and whether a SQUID feedback loop could keep up with a signal that fast.

Those objections leave the simulated numbers intact and question only how far they can be pushed toward a working device. The published version, listed in Physical Review B volume 114, issue 8 under DOI 10.1103/6gkt-hghj, says the readout approaches the nondestructive limit arbitrarily closely, a softer claim than the 2024 preprint’s suggestion that the qubit’s state is left intact. Some coverage called the venue Physical Review B Letters; the DOI listing and the reference line say Physical Review B.

Open problems before a lab can test the SQUID scheme

Egues is direct about the limits. He said the method does not eliminate the fundamental problem of quantum measurement, and that it can only approach a nondestructive readout. He also said every parameter used in the simulations is realistic and drawn from existing devices, and that in principle the result can be reached. What is missing is the particular geometry and the assembled apparatus.

The detector itself is a second catch. The SQUID has to sit close enough to sense the oscillating flux, and in doing so it feeds a small current back onto the qubit. The team argues that this feedback is negligible, and its simulations treat it that way, but the claim remains to be confirmed on hardware. Reporting by The Quantum Insider repeats the same list of caveats: purely theoretical, geometry not yet built, feedback reduced but not removed.

Nobody has yet shown the effect in a laboratory. An experiment would have to show the predicted oscillating supercurrent in a real indium arsenide nanowire device and then reconstruct a known state from it at the quoted accuracy. Until a group does that, the 99.85% remains a simulated figure for one detection scheme, and the open question is whether the SQUID can follow a qubit oscillating in the gigahertz range without drowning the signal in its own back-action.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

The day in tech, in one email: Get TLDR →

We may earn a commission if you sign up.