Quantum Computing Moves Toward Reality.
But until recently, that possibility has remained only theoretical, as no one has known how to build a quantum computer, much less program one. That seems to be shifting now, with a cascade of advances, not only in building the quantum bits but also in quantum memory devices and processes to link the qubits together into a working computer.
The theory behind quantum computers has been around for more than 30 years. It originated in a 1982 paper in which visionary physicist Richard Feynman reasoned that since the actions of subatomic particles are predictable, at least in large numbers, their behavior could be seen as a sort of calculation and mused about the possibilities. A decade later, in 1994, Bell Labs researcher Peter Shor published a paper showing that a theoretical quantum computer could factor prime numbers orders of magnitude faster than traditional computers. Factoring primes is at the heart of modern encryption and computer security; Shore demonstrated that a quantum computer could break that encryption in a matter of hours. At the time, the threat was theoretical--neither quantum computers nor the programming that would drive them yet existed.
In recent years, however, university researchers around the globe have reported startling advances in quantum computing. Since the first experimental quantum logic gate was constructed in 1995, the science has taken off. The first two- and three-qubit quantum computers solved their first problems in 1998, with five- and seven-qubit computers coming two years later. Today, research has moved beyond university and government research labs. More than a dozen companies are investing in building quantum computers, including Intel, IBM, and Google; IBM and Google are already making their first quantum computers available to researchers via the cloud.
The problem of the quantum computer is not solved, however; a number of challenges remain. One of the longest-standing problems is that qubits are extremely fragile; in current models, they operate at temperatures near absolute zero because virtually any vibrations knock them out of their quantum state, producing errors. For qubits to be useful in computing, they must achieve both quantum superposition (occupying both states at once) and entanglement (twinned with another particle so that the two affect each other instantly no matter how far apart---what Einstein called "spooky action at a distance"). Even the slightest vibration or variation in charge kicks them out of this quasi-magical condition. And the complexity of the problem grows with the number of connections.
Nor are all qubits the same. There are two basic types: spin qubits rely on one of the peculiar, impossible-to-visualize properties of subatomic particles; superconducting qubits are built around a circuit called a Josephson junction, a tiny insulator placed between superconducting wires, named for Cambridge physicist Brian Josephson.
There are three kinds of superconducting qubits--flux, charge, and phase qubits; all use Josephson junctions. Superconducting loops that include a Josephson junction allow current to flow clockwise and counterclockwise simultaneously, a superposition of states. Manipulating the current through the wires allows physicists to set up qubits in these systems. One advantage of superconducting qubits is that they can be made on silicon, using some of the same techniques developed for ordinary IC chips.
Another approach to quantum computing manipulates the spin of single atoms. Spin is a quality of subatomic particles that makes them appear to act like tiny magnets in some experiments. Spin can be either "up" or "down" and the particle's "spin number" is used in physics equations, but the terms are only metaphorical; nothing is spinning, in the normal sense, and up and down are not directions that can be pointed to.
Researcher Andrea Morello is leading a team at the University of New South Wales, in Sydney, Australia, that is making spin qubits. The team has been able to make a phosphorus atom trapped in a silicon device hold two spin values at the same time, then to read the values by applying a microwave pulse. In the single-atom qubit used by Morello's team, a silicon chip is covered with a layer of insulating silicon oxide with a pattern of metallic electrodes on top, operating near absolute zero and in the presence of a strong magnetic field. While most qubits only maintain their quantum state for a matter of milliseconds, these devices have maintained coherence for several seconds--a vast amount of time in quantum computing terms.
A member of Morello's team, Guilherme Tosi, has also created a third, entirely new type of qubit that uses both the nucleus and the electron of the atom. Instead of having one particle held in two states at once, in Tosi's qubit, the electron is held in one state and the atom's nucleus in the other. Flipping their states changes the value of the qubit between 0 and 1. The "flip-flop qubits," as Tosi calls them, can be spaced much farther apart than spin-type qubits while remaining entangled, leaving room for traditional computer components like interconnects and readout devices and making it easier to build quantum computers with thousands or millions of qubits.
Still other researchers are developing quantum memory devices and ways to network groups of quantum computers. In a demonstration of a part of a quantum memory device, scientists at Tsinghua University in Beijing made a qubit consisting of a magnetically trapped, positively charged ytterbium-171 ion, which achieved a coherence time of more than 10 minutes.
Possibly the strongest sign that quantum computing has moved beyond vaporware is the fact that some of the largest and most successful computer companies in the world are in a competition to build functional quantum computers. Intel, IBM, and Google have all hired physicists working in the field of quantum computing and building their own devices. On the hardware side, Google researchers in Goleta, California, have said that they hope to build a 7-by-7 array of superconducting qubits on a quantum IC by the end of 2017 or early in 2018, having already made six- and nine-qubit chips. Although 49 qubits are far fewer than what scientists estimate will be needed to fuel the truly amazing feats quantum computing is thought to be capable of, it is the number the Google team believes is needed to demonstrate "quantum supremacy"--proof that quantum computers really can outstrip conventional supercomputers in solving actual problems.
Other companies are also in the race. In October, Intel announced it had shipped to research partners in Europe an experimental quantum computing chip that held 17 super-conducting qubits. The number 17 is significant, according to Intel's director of quantum hardware--it is the minimum number of qubits needed to run a kind of code error correction algorithm thought to be necessary for scaling quantum computers to useful sizes.
And IBM has simulated a 56-qubit quantum computer on a traditional supercomputer. The simulation allows programs like the error correction algorithm to be tested before a fully functional 56-qubit machine can be built. It can also be used to predict how the quantum device should function and help locate faults when one is built. At the same time, the IBM team is making a five-qubit device with a drag-and-drop interface available to average users in the cloud. The company hopes to learn what does and, especially, what doesn't work from people's engagement with the device.
Quantum computing promises to revolutionize computing in ways that will trickle down into daily life. Large-scale quantum computers will almost certainly lead to truly uncrack-able computer security. Quantum computers made up of a million or more qubits could revolutionize chemistry and drug development by accurately modeling molecules at the subatomic level, a feat that would require a classical supercomputer with a number of bits equal to all the atoms on planet Earth. They may even answer questions like whether the brain is actually a quantum computer, as some researchers have suggested. And that future is coming, sooner than many expect. Experts predict we may see commercial million-qubit computers in as little as 5 or 10 years. The change could turn out to be as momentous as trading up from an abacus to a smartphone.
Manny Frishberg, Contributing Editor
Federal Way, Washington
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||News and Analysis of the Global Innovation Scene|
|Date:||Mar 1, 2018|
|Previous Article:||A Cloud of Hope for Shared Science.|
|Next Article:||Innovation Flashback.|