Quantum computers should theoretically outpace ordinary ones, but attempts to build a speedy quantum machine have so far come up short. Now an approach based on a Victorian counting device seems to be getting close.

This proto-quantum computer can only solve one problem. But that problem, called boson sampling, seems to be very difficult for an ordinary computer to solve, so physicists hope that such a device will conclusively demonstrate the promise of computing based on exotic physics. “The goal is to show quantum supremacy with the simplest approach,” says Fabio Sciarrino of Sapienza University in Rome, Italy, who helped develop the new machine.

Boson samplers are based on a device created by the 19th-century polymath Francis Galton to study statistical distributions. It consists of a wooden board studded with offset rows of pegs. Balls are dropped one by one from the top of the board and ping their way down, bouncing left or right at each peg, before collecting in bins at the bottom. Since balls are more likely to end up in a central bin than one at the edges, you end up with a bell curve distribution across the width of the board. In the pre-computer age, it was one of the best ways to compute this distribution, which often crops up in statistics.

The quantum version swaps balls for photons, which travel along a network of intersecting channels in an optical chip. When two photons collide, their ensuing paths are determined by the laws of quantum mechanics, producing a unique distribution. With enough photons, calculating this distribution becomes very difficult on an ordinary computer, so doing it with real photons in a quantum device is the only practical option.

In 2012, four research groups, including Sciarrino’s, demonstrated the first working boson samplers with three photons. But scaling to larger numbers was challenging as it is difficult to produce single photons on demand. The leading method, which involves shooting a laser at a crystal, spits out photons at random times, so you can’t get enough in the boson sampler at once.

### Scattershot approach

That’s why Sciarrino has turned to a slightly different version of the problem, called scattershot boson sampling. This involves using a larger number of photon sources, so that their randomly generated photons have a higher chance of colliding. His team used six sources and were able to produce three photons at once. That means the new boson sampler is no more computationally powerful than the 2012 examples, but operates on average 4.5 times as fast.

“We only have shown a proof-of-principle improvement,” says Sciarrino. Now his team is working to improve its sources even further, with the aim of challenging an ordinary computer – each additional photon roughly doubles the difficulty of the calculation. “If you want to start seeing a large improvement, you need something like 20 or 30 sources.”

Scott Aaronson of the Massachusetts Institute of Technology, who helped come up with the idea of boson sampling, thinks the device is an important milestone, but not yet a breakthrough. “Hopefully they will have better scaling going forward,” he says. “Twenty or 30 photons would be spectacular.”

A boson sampler isn’t inherently useful, although last year a group of researchers at Harvard University published a theoretical paper suggesting one might be able to calculate the vibrational properties of certain molecules. In any case, the skills and technology needed to get one working with more photons should help enable more general-purpose quantum computers in the future, says Aaronson.

Even if it can’t factor large numbers or perform other quantum tricks, a device that unambiguously demonstrates quantum supremacy would be a major scientific breakthrough. Perhaps the first record-beating boson sampler will one day sit in a museum alongside Charles Babbage’s difference engine, the mechanical precursor to modern computers. “I like that image,” Aaronson says. “I’d go visit it in a museum.”

Journal reference: *Science Advances*, DOI: 10.1126/sciadv.1400255

If you would like **to reuse any content** from New Scientist, either in print or online, please **contact the syndication** department first for permission. New Scientist does not own rights to photos, but there are a variety of licensing options available for use of articles and graphics we own the copyright to.