## First universal programmable quantum computer unveiled

发布时间：2017-10-13 06:01:15来源：未知点击：

By Colin Barras The world’s first universal programmable quantum computer has been put through its paces. But the test program revealed significant hurdles that must be overcome before the device is ready for real work. Earlier in the year, a team at the National Institute of Standards and Technology in Boulder, Colorado, built a quantum computer capable of processing two quantum bits, or qubits. Qubits store more information than the simple “on” or “off” bits of conventional computing, which means that a quantum computer outperform conventional computers in tasks such as cryptanalysis. As in a classical computer, a series of logic gates processes the information – although here the gates are quantum logic, or qubit, gates. “For example, a simple single-qubit gate would change a ‘one’ to a ‘zero’ and vice versa,” says David Hanneke, a member of the team. But unlike the physical logic gates of a classical computer, the quantum logic gates used in the team’s device are each encoded into a laser pulse. The experimental device uses beryllium ions to store qubits in the way they spin while the laser-pulse quantum gates perform simple logic operations on the qubits. The trick to making a quantum logic gate is in designing a series of laser pulses that manipulate the beryllium ions in a way that processes information. Another laser then reads off the results of the calculations. “Once we had demonstrated we could successfully combine lots of components in this way, we ask: what can you do with that?” says Hanneke. They found their answer in quantum computational theory. “One of the more interesting results to come out of the early years of quantum information was that you can do any quantum operation on any number of qubits using only single and two-qubit logic gates,” says Hanneke. Although one and two-qubit gates have already been built and used to perform specific algorithms, no one had yet built a device capable of all possible quantum routines. Until now. At the heart of the device is a gold-patterned aluminium wafer containing a tiny electromagnetic trap some 200 micrometres across, into which the team placed four ions – two of magnesium and two of beryllium. The magnesium ions act as “refrigerants”, removing unwanted vibrations from the ion chain and so keeping the device stable. There are an infinite number of possible two-qubit operations, so the team chose a random selection of 160 to demonstrate the universality of the processor. Each operation involves hitting the two qubits with 31 distinct quantum gates encoded into the laser pulses. The majority were single-qubit gates, and so the pulse needed to interact with just one ion, but a small number were two-qubit gates requiring the pulse to “talk” to both ions. By controlling the voltage on the gold electrodes surrounding the trap, the team can uncouple the ions when single-qubit gates are needed and couple them again for two-qubit operations. The team ran each of the 160 programs 900 times. By comparing the results with theoretical predictions, they were able to show that the processor had worked as planned. But it did so with an accuracy of only 79 per cent, says Hanekke. “Each gate is more than 90 per cent accurate, but when you stack them together the total figure falls to 79 per cent or so for a given operation,” he says. That’s because each of the laser pulses that act as the gates varies slightly in intensity. “They’re not ‘square’ pulses [that switch on and off cleanly] – they fluctuate,” he says. And the beam has to be split, reflected and manipulated in various ways beforehand, which also introduces errors. Such errors would drown the results of any more extensive computations. The fidelity needs to increase to around 99.99 per cent before it could be a useful component of a quantum computer. That could be done by improving the stability of the laser and reducing the errors from optical hardware, says the team. If those levels of accuracy can be reached, the new chip could form an integral part of a useful quantum processor. “If you have a simple and repetitive task you might have a dedicated region [of the processor] to do that,” he says. “But you need regions that can do all kinds of stuff – this is just such a device.” Journal reference: Nature Physics, DOI: 10.1038/nphys1453 More on these topics: