Two quantum computers go head-to-head for the first time

IBM's device is faster but the University of Maryland's machine is more accurate

Two prototype quantum computers have gone head-to-head for the first time in a challenge to see which is the fastest and most accurate.

The two devices, one created by IBM and the other by the University of Maryland, work in different ways, but are able to compete on a like-for-like basis because they process algorithms in the same way.

Both machines are five-qubit devices, the qubit being the unit of quantum information. The IBM device is based around five loops of superconducting metal, the currents through which are controlled by microwaves, whereas the University of Maryland machine uses five ytterbium ions constrained by an electromagnetic field and manipulated using lasers.

The experiment, which was led by Chris Monroe, a physicist at the University of Maryland, was made possible because IBM opened up its quantum chip to make it programmable by regular online users, rather than just by scientists in the lab, as has been the case up until now.

The difficulty in maintaining the superposition states for the duration of a calculation as well as the probabalistic nature of quantum mechanics means that, as yet, quantum computers are not nearly as accurate as their digital counterparts and therefore cannot be considered general purpose devices. But the fact they can perform calculations in parallel rather than sequentially makes them many times faster.

During the experiment, a set of standard algorithms was run on each device and the output compared. In the event, the University of Maryland computer was more accurate, getting the answer right an average of 77.1 per cent times across all of the algorithms tested. IBM's superconducting loops could only achieve a success rate of 35.1 per cent. However, it was around 1,000 times faster than the ytterbium ion machine.

The reason for the discrepancies lies in the fragility of the quantum states of the interconnected qubits. In Monroe's device, each ion is able to interact with all the others leading to greater stability, whereas IBMs superconducting loops interact via a central loop, making the quantum state more prone to collapse. However, IBM believes that the stability of qubit superposition will improve and become more coherent over time. Indeed, Jerry Chow of IBM's quantum computing team has stated that the latest version of IBM's device now features more qubit connections than it did when the experiment was run, bring it closer to its rival in the stability stakes.

The two devices are a long way from being useful general purpose computers. A practical device will require thousands of interacting qubits, the challenge being that maintaining a stable state becomes much more difficult with each qubit added. However, the experiment is a sign of progress according to Simon Benjamin, a physicist at the University of Oxford.

"For a long time, the devices were so immature that you couldn't really put two five-qubit gadgets next to each other and perform this kind of comparison, " he told Science. "It's a sign that this technology is maturing."