Media release

From: Springer Nature

Quantum computing: Steps towards potentially useful near-term future quantum processing (N&V)

The demonstration of a quantum processor that outperforms classical computations without the implementation of error correction is reported in Nature this week. An IBM 127-qubit processor is shown to prepare and measure expectation values (an estimated average result of repeated experiments) of highly entangled quantum states that are beyond the capabilities of current best classical computational methods. The demonstration suggests that quantum processors may potentially be useful for some specific computations in the near future even without fault-tolerance — the operation of a quantum computer where errors are avoided or corrected quickly enough to be under control — whereas fault-tolerant computing is likely to take many more years to be achieved.

A key goal of quantum computing is to perform specific tasks more efficiently than what is possible with classical computers. To achieve this goal, a number of practical challenges need to be addressed, such as keeping error rates low and cutting through quantum ‘noise’ (disturbances from the underlying system or environment) while increasing the size of the quantum computer. Errors and noise reduce or erase any advantage that quantum computing may offer over classical calculations. Fault-tolerance remains out of reach for existing technologies. Although existing quantum processors have been shown to outperform classical machines on specific but contrived problems, it has been debated whether current or near-future noisy quantum computers may be good enough to execute quantum computations that could be of use, for instance, for research purposes.

Andrew Eddins, Youngseok Kim, Abhinav Kadala and colleagues provide evidence that their quantum chip can reliably generate, manipulate, and measure quantum states that are so complex that their properties cannot be reliably estimated by classical approximations. This demonstration suggests that quantum machines may be already able to help with some specific problems — such as studying physics models — which are intractable on classical computers, even without error correction. The authors report experiments on a 127-qubit processor running circuits 60 layers deep with around 2,800 two-qubit gates (the quantum equivalent of classical computer logic gates). Such a quantum circuit generates large and highly entangled quantum states, which are too demanding to be reliably reproduced by numerical approximations on a classical computer. The authors show that their quantum computer could instead accurately estimate the properties of these states by measuring expectation values. Creating and measuring such large states without generating so many errors as to undermine the computation was enabled by the high quality of the fabricated chip and by a post-analysis processing method that compensates for noise.

“The fundamental quantum advantage here is the scale rather than speed — the 127 qubits encode a problem in a huge state-space for which no classical computer has enough memory,” note Göran Wendin and Jonas Bylander in an accompanying News & Views.