By Jacob Aron A quantum version of a Victorian counting machine could be just what quantum computers need to race ahead of the pack. Quantum computers have long promised to beat ordinary hardware when it comes to quick number crunching, but so far no one has managed to create one that can beat even a pocket calculator. Last year, Scott Aaronson of the Massachusetts Institute of Technology suggested tackling this problem by rewinding the technological clock. His idea was to build a quantum version of a Galton board, invented by Victorian scientist Francis Galton as a way to study statistical distributions. The device is an upright board studded with a symmetrical arrangement of pegs. The pegs scatter balls dropped in at the top, causing them to land in different slots at the bottom. The arrangement is designed so that balls are most likely to land in the central slot and least likely to land in the slots at the edges. If enough balls are dropped in, the pattern – or sample – they form at the bottom approximates a binomial distribution, a key concept in statistics. Galton’s board isn’t exactly a computer, but it is clearly able to calculate something. “If you were living in the pre-digital-computer era, it would actually be an extremely convenient way to obtain such samples,” says Aaronson. He suggested that it should be possible to construct a quantum equivalent to the board – a device that uses the laws of quantum, rather than classical, mechanics to perform certain calculations – and that such a device would carry out these calculations faster than a machine using any other method, including conventional computation. Now a number of teams have built such a device, swapping the balls for photons. Their “boson samplers”, named after the family of particles to which photons belong, consist of a number of channels that snake from one end of a board to the other. The channels intersect at certain points, allowing photons from different channels to interfere and change their paths. When photons are injected into certain tunnel openings at one end according to a pattern, their emergence at the other follows a distribution that is predictable – similar to the binomial distribution produced by the pegboard. Crucially, though, their paths are governed by the laws of quantum mechanics, ensuring that macroscopic balls sent through a much larger version of the device would not produce the same distribution. Unlike the balls, which can go left or right with equal probability as they hit each peg, quantum effects mean that when two photons within the network meet, they must both go left or both go right. “It’s that little bit of physics which makes this machine work in a quantum way,” says Ian Walmsley of the University of Oxford, who led one of the teams behind the boson sampler. Once the patterns get sufficiently complicated, producing the same distribution in a non-quantum way would require much more computation and so take much longer. So far, though, no team has sent more than three photons at a time through their boson samplers, not enough to show a performance boost over classical computing. But building a larger device should be much easier than attempting to scale up true quantum computers, say the researchers. If the boson sampler can be scaled up, it will serve as the first instance of a quantum mechanism carrying out a calculation faster than any classical mechanism. “This really is a place where a quantum machine can significantly outperform a classical one,” says Walmsley. Like the Galton board, the boson samplers that have been made are not genuine computers because they can only carry out this particular type of calculation. A true quantum computer would, in principle, be capable of any calculation. It is theoretically possible to turn the boson sampler into a universal quantum computer by adding in additional ways to manipulate the photons as they pass through, but this is extremely difficult to achieve without messing up the device, says Aaronson. Walmsley’s team, as well as one led by Matthew Broome of the University of Queensland, Australia, which Aaronson was part of, have just published their work in the journal Science, while two other groups have posted preliminary papers describing similar experiments. Journal references: Science, DOI:10.1126/science.1231692 and DOI:10.1126/science.1231440, plus http://arxiv.org/abs/1212.2240 and http://arxiv.org/abs/1212.2783 More on these topics: