Practical Quantum Computers Creep Closer to Reality

Physicists find quantum versions of both feedback control and classical computer architecture

3 min read

16 September 2011—The long-promised arrival of practical quantum computers—machines that exploit the laws of quantum mechanics to solve complex problems much faster than conventional computers do—seems a step closer, thanks to two recent advances by physicists.

In the first development, reported in the 2 September issue of Nature by a group led by Serge Haroche of the École Normale Supérieure and the Collège de France in Paris, the researchers created a real-time feedback mechanism for a quantum computer. Control mechanisms, such as feedback loops, are central to the operation of large conventional computers.

In the second advance, reported the same week in Science by a group led by Matteo Mariantoni and John Martinis of the University of California, Santa Barbara, scientists created a quantum central processing unit (CPU) with memory. The rudimentary device is the first quantum computer based on the common von Neumann processor-memory architecture that conventional computers use.

Dick Slusher, director of the Quantum Institute at the Georgia Institute of Technology, in Atlanta, and other experts unanimously praised the work of both groups. However, Slusher says that ”for quantum computing to be fault tolerant—a condition required to scale up to true applications like factoring useful coding keys—the error levels must be much lower than achieved so far.”

Quantum computing is an emerging field that has witnessed considerable advances in recent years, including progress toward silicon devices. However, it has proved difficult to create a practical quantum computer that would rival the processing abilities of a conventional machine. Part of the difficulty lies in the fragility of quantum states, which break down (or ”decohere,” in the parlance of quantum mechanics) rather quickly. So far, only rudimentary quantum computers with a handful of ”qubits” (quantum bits) have been built. (In May, D-Wave Systems sold Lockheed Martin a special type of computer that relies on a ”quantum annealing” processor, but many quantum computing experts remain skeptical that it is a true quantum computer.)

As they seek to create larger quantum systems, scientists have tried to incorporate some of the same systems-engineering concepts that are used in conventional computers, but the equivalent quantum systems have proved elusive—until now. ”These machines are very fragile,” says Haroche. ”The coupling to their environment causes decoherence, which destroys the quantum features required to achieve their tasks. Correcting the effects of decoherence is thus a very important aspect of quantum information. One possibility is to control the quantum machine by quantum feedback.”

Yet therein lies a challenge: In the quantum world, the mere act of observing photons or atoms perturbs their motion and changes their positions and velocities—and therefore the value the qubit holds. So for quantum feedback to work, one must be able to observe the system by performing ”weak measurements,” perturbing it only minimally, and the computer must take the perturbation into account before applying the correction.

Haroche and his colleagues use a small collection of atoms as a kind of quantum sensor to overcome this challenge. They pass atoms through a microwave cavity that contains the qubits as photons. The atoms obtain a detectable signal—a shift in their phase. This technique provides information about the state of the photons, but it does so by performing only a weak measurement and does not lead to a total collapse of the light’s quantum nature. Measuring changes in the final state of atoms that sequentially pass through the light field provides a signal that can be used to control the light.

”The work is a very impressive demonstration experiment showing that the many techniques developed in the systems engineering community can be translated to the quantum regime—if one is clever enough,” says Michael Biercuk, a quantum physicist at the University of Sydney, in Australia.

The challenge of translating a classical system, in this case the common von Neumann processor-memory architecture, into a quantum system also motivated the second team of researchers. To build a quantum CPU and RAM, the UC Santa Barbara group used two superconducting Josephson junctions—two pieces of superconducting metal separated by a thin insulating layer—as qubits. They connected the qubits using a bus made of a superconducting microwave resonator. Each qubit also had a separate resonator that acted as RAM. With the help of microwave pulses, the qubits could influence one another’s state in a way that performed calculations, and the results could be stored in the quantum RAM. They tested their CPU by allowing it to solve a few quantum algorithms, including the equivalent of the Fourier transform. The demonstration could quickly lead to a larger-scale quantum processor based on superconducting circuits, according to the UC Santa Barbara team.

The most complex algorithms performed so far have used a quantum computing system based on trapped ions, but Biercuk says the superconducting system is quickly catching up, and that’s ”extremely exciting.”

While no one expects a quantum computer to rival a conventional computer in the very near future, experts were pleased with these recent developments.

Raymond Laflamme, executive director of the Institute for Quantum Computing at the University of Waterloo, in Canada, said both experiments had ”very strong results,” and that they ”demonstrate an increasing amount of control of quantum processors.”

About the Author

Saswato R. Das, a New York City–based writer, contributes frequently to IEEE Spectrum. For one assignment, Das got the last interview with famed science fiction writer Arthur C. Clarke before he died in 2008.

The Conversation (0)

The Future of Deep Learning Is Photonic

Computing with light could slash the energy needs of neural networks

10 min read

This computer rendering depicts the pattern on a photonic chip that the author and his colleagues have devised for performing neural-network calculations using light.

Alexander Sludds

Think of the many tasks to which computers are being applied that in the not-so-distant past required human intuition. Computers routinely identify objects in images, transcribe speech, translate between languages, diagnose medical conditions, play complex games, and drive cars.

The technique that has empowered these stunning developments is called deep learning, a term that refers to mathematical models known as artificial neural networks. Deep learning is a subfield of machine learning, a branch of computer science based on fitting complex models to data.

Keep Reading ↓ Show less