Google’s Sycamore quantum computer Rocco Ceselin/Google
Google has shown that its Sycamore quantum machine tin observe and hole computational errors, an indispensable measurement for large-scale quantum computing, but its existent system generates much errors than it solves.
Error-correction is simply a modular diagnostic for ordinary, oregon classical, computers, which store information utilizing bits with 2 imaginable states: 0 and 1. Transmitting information with other “parity bits” that pass if a 0 has flipped to 1, oregon vice versa, means specified errors tin beryllium recovered and fixed.
In quantum computing the occupation is acold much analyzable arsenic each quantum bit, oregon qubit, exists successful a mixed authorities of 0 and 1, and immoderate effort to measurement them straight destroys the data. One longstanding theoretical solution to this has been to clump galore carnal qubits into a azygous “logical qubit”. Although specified logical qubits person been created previously, they hadn’t been utilized for mistake correction until now.
Julian Kelly astatine Google AI Quantum and his colleagues person demonstrated the conception connected Google’s Sycamore quantum computer, with logical qubits ranging successful size from 5 to 21 carnal qubits, and recovered that logical qubit mistake rates dropped exponentially for each further carnal qubit. The squad was capable to marque cautious measurements of the other qubits that didn’t illness their authorities but, erstwhile taken collectively, inactive gave capable accusation to deduce whether errors had occurred.
Kelly says that this means it is imaginable to make practical, reliable quantum computers successful the future. “This is fundamentally our archetypal fractional measurement on the way to show that,” helium says. “A viable mode of getting to truly large-scale, error-tolerant computers. It’s benignant of a look up for the devices that we privation to marque successful the future.”
The squad has managed to show this solution conceptually but a immense engineering situation remains. Adding much qubits to each logical qubit brings its ain problems arsenic each carnal qubit is itself susceptible to errors. The accidental of a logical qubit encountering an mistake rises arsenic the fig of qubits wrong it increases.
There is simply a breakeven constituent successful this process, known arsenic the threshold, wherever the mistake correction features drawback much problems than the summation successful qubits bring. Crucially, Google’s mistake correction doesn’t yet conscionable the threshold. To bash truthful volition necessitate little noisy carnal qubits that brushwood less errors and larger numbers of them devoted to each logical qubit. The squad believes that mature quantum computers volition request 1000 qubits to marque each logical qubit – Sycamore presently has conscionable 54 carnal qubits.
Peter Knight astatine Imperial College London says Google’s probe is advancement towards thing indispensable for aboriginal quantum computers. “If we couldn’t bash this we’re not gonna person a ample standard machine,” helium says. “I applaud the information they’ve done it, simply due to the fact that without this, without this advance, you volition inactive person uncertainty astir whether the roadmap towards responsibility tolerance was feasible. They removed those doubts.”
But helium says it volition beryllium a immense engineering situation to really conscionable the threshold and physique effectual mistake correction, which would mean gathering a processor with galore much qubits than has been demonstrated until now.
Journal reference: Nature, DOI: 10.1038/s41586-021-03588-y
More connected these topics:
English (US) ·