news center

Controversial quantum computer aces entanglement tests

作者：吉觉 时间：2019-03-13 10:12:12 人气： ℃

By Jacob Aron Editorial: “Quantum computers leap into the real world“ A few days ago, I held a quantum computer in my hand – or did I? The niobium wafers I saw are the guts of the only commercially available quantum computer, but whether their calculations truly harness the weird world of quantum mechanics has provoked heated debate. Now the chips, made by D-Wave of Burnaby, Canada, have passed two tests that suggest that the bits in their machines have the quantum property of entanglement. That doesn’t end the controversy, but it strengthens D-Wave’s claim that a revolution in computing is a lot closer than we thought. Though D-Wave has been selling its chips (see picture) since 2011, its quantum computing credentials have remained controversial. That’s because the firm’s devices go against the traditional model of quantum computing, which is based on quantum versions of the logic gates found in ordinary, or classical, computers. These gates manipulate qubits, which, unlike classical bits, can represent both a 0 and 1 at the same time, making it theoretically possible to compute the answers to large problems much faster than on an ordinary computer. However, these gate-based systems are nowhere near being commercially available because the largest of them consists of just a few bits, not nearly enough for these theoretical gains to be realised. D-Wave instead uses adiabatic quantum computing, in which an array of chilled, superconducting niobium loops – the qubits in this system – very quickly find the lowest point in what can be thought of as an energy “landscape” of hills and valleys. The trick is to use this landscape to answer a variety of questions. For example, D-wave has arranged the qubits so that when they reach their lowest-energy state , they reveal the lowest-energy arrangement for folding a simple protein – thought to be its preferred state. To find this lowest-energy “ground state”, the qubits exploit a quantum effect called tunnelling: they go through, rather than over, the hills. That allows them to explore the energy of all the possible states much faster than a classical system, which would need to climb each hill. In 2011, D-Wave published evidence that its qubits were indeed tunnelling, but to count as a quantum computer, capable of realising the theoretical speed gains that come with it, the qubits must also be entangled – linked by quantum mechanics, no matter how far apart they are in space. Entanglement cannot be directly measured while the D-Wave computer is operating. However, it will have an effect on the qubits’ energy distribution. So D-Wave has developed a technique called qubit tunnelling spectroscopy that measures the energies of the qubits and uses this to determine whether it corresponds to what you would expect from an entangled system. On 6 March, at the Adiabatic Quantum Computing workshop at the Institute of Physics in London, they reported that they had found this entanglement signature in experiments with systems of two and eight qubits within their computer. The indirect nature of the measurement means the calculations rely on an assumption about the energy spectrum – which could turn out to be wrong, invalidating that conclusion. But D-Wave’s results aren’t the only evidence that their system is entangled. At the same workshop, Federico Spedalieri of the Information Sciences Institute in Marina Del Rey, California, presented additional evidence of entanglement, using data provided by D-Wave but employing a different methodology. Spedalieri and colleagues applied a mathematical test that determines whether there are any ways for non-entangled qubits to arrange themselves to be compatible with the data. If not, the system must be entangled. Using this test, they found evidence for entanglement. What’s more, it was at a similar stage in the computation to where D-Wave also found evidence for entanglement, based on energy distribution. “We’re very confident that after several microseconds there is entanglement,” says Spedalieri. “With this evidence it would be a very conspiratorial set of parameters that will make this non-entangled.” The ISI also has its own D-Wave computer which it operates on behalf of defence firm Lockheed Martin, D-Wave’s first paying client. Entanglement doesn’t mean case closed. “It is a necessary, but not sufficient, condition,” says Aram Harrow of the Massachusetts Institute of Technology. To have a true quantum computer, D-Wave now needs to demonstrate that the entanglement actually translates into performance gains. For traditional quantum computing, theory shows that scaling up the computer’s size translates into exponential speed gains. But adiabatic computing is a much newer field and so much less of the theory is known. Although some speed-up should occur in an ideal, quantum system at absolute zero, no one knows whether this can be realised in D-Wave’s processors, which run at just above absolute zero. “Nor are there negative results proving it can’t be done,” adds Harrow. “It is an open question.” And if it can indeed be done, the size of the speed gain as the chips are scaled up is itself a mystery. D-Wave claims that its newest 512-qubit processor, which should come to market this year, demonstrates a significant speed-up when compared to the leading classical algorithms – as much as 10,000 times faster for problems requiring a large number of qubits. The company plans to publish the results later this year. Repeat experiments are also still needed to confirm entanglement beyond statistical doubt. In the meantime, the workshop response to the results seemed positive. “As time has gone on, we have accumulated more and more evidence of quantum-mechanical behaviour,” says Williams. “It will be conclusive very soon that the hardware is quantum.” Computers using D-Wave’s devices will always differ from gate-based quantum computers in one key way: D-Wave’s devices do not lead to universal computers that can be programmed with any algorithm, as an ordinary computer can. Instead, says D-Wave’s Colin Williams, their chips focus on solving a particular class of optimisation problems. “That’s such an important class that commercially it is a much better strategy than trying to make a universal machine.” More on these topics: