NIST physicist David Wineland adjusts an ultraviolet laser beam used to manipulate ions in a high-vacuum apparatus containing an "ion trap." These devices have been used to demonstrate the basic operations required for a quantum computer. Such computers, by relying on quantum mechanics rather than transistors to perform calculations or store information, could someday solve problems in seconds that would take months on today's best supercomputers. (Geoffrey Wheeler / NIST)
Quantum computing research is now officially Nobel Prize-worthy. American physicist David Wineland and French physicist Serge Haroche traveled to Stockholm in early December to share the 2012 physics honors for work their teams have done separately on the subject.
For digital natives accustomed to thinking of smaller as better, the first quantum computers won’t look prize-worthy. If and when they hum to life, they’ll be at least as big as refrigerators or modern-day supercomputers. A quantum computer might even take up an entire room. It would almost certainly need old-fashioned classic computers to manage the strange quantum dance generated by lasers, cryogenic coolers and exotic materials. But if scientists can precisely choreograph that tango, quantum computers will offer exponentially more computation power than today’s machines.
Not surprisingly, that potential has caught the collective eyes of the intelligence community. A full-scale version might reduce most existing cryptography to arithmetic.
Although Wineland and Haroche are being honored for pioneering work they did in the 1990s, experts estimate that a working quantum computer remains 10 to 15 years away. There won’t be many of them, at least at first.
“The government might have one or two,” predicts mathematician Dickie George, who retired from the National Security Agency last year after a long career as the agency’s top cryptographer.
Still, there is an international race underway. Russia, China and others are all trying to harness quantum’s potential, and the U.S. is determined to solve the puzzle first. To that end, Washington is underwriting millions of dollars of research each year.
At stake is an almost unimaginable power. As Wineland explains it, quantum computers would have more unique combinations of bits at their disposal than there are “elementary particles in the universe.”
Here’s why: Most cryptography is based on public-key infrastructure (PKI), a 35-year-old encryption and decryption technique that underlies the secure electronic communications of law enforcement, government agencies, financial institutions, and even billions of dollars of consumer e-commerce. To crack such systems, a code breaker would need to compute the lengthy prime numbers that are mathematical factors of huge numbers. That’s been a daunting challenge — so far.
“Quantum computing is going to blow those systems out of the water,” a retired intelligence official said.
In part, that’s because the math required to factor huge numbers is well-understood. An algorithm to do so was drafted in 1994 by U.S. mathematician Peter Shor. All that’s needed now is a computer powerful enough to run the algorithm — a quantum computer.
Once it appears, just about the last bastion of security would be the National Security Agency’s most secure “black” encryption codes. Those keys are longer than the public keys, and they are kept private on the encryption side and decryption side. But these codes are cumbersome to use. Break PKI, and in theory, at least, many of the world’s secrets could be exposed.
The stakes are high enough that quantum researchers don’t expect the funding spigot to be closed, even with today’s budget pressures.
“I think it’s so key to the intelligence community’s business that I’d be very surprised if it goes away,” said Wineland.
Wineland, who does his research at the National Institute of Standards and Technology lab in Boulder, Colo., revealed that several intelligence agencies have been funding his research since 1995. The list includes the Intelligence Advanced Research Projects Activity, the Defense Advanced Research Projects Agency, the Army Research Office, the Office of Naval Research and other agencies Wineland said he’s not permitted to name.
Wineland is competing with researchers around the world, including some in Russia and China, in a chase for a new way to run calculations.
Classic computers store and crunch information by turning on or off the current to individual circuits etched in silicon chips. The number 1 is traditionally represented by letting five volts flow through. The number 0 is represented by keeping the circuit off. Each year, companies such as Intel, AMD and Nvidia figure out how to squeeze more circuits on individual microchips. Engineers know that sooner or later, they’re going to run out of room. Eventually, Moore’s Law, which holds that the number of circuits squeezed onto a microchip doubles about every two years, will collapse.
Quantum computing would open a new dimension. It would mean harnessing the mysterious quantum properties of individual electrons, atoms or photons — the building blocks of matter and energy. The specific attribute that is so promising is this: In certain conditions, quantum particles can be in two places at the same time, which means they can simultaneously exhibit different states of energy. It’s a fascinating quality, difficult even to comprehend — and it has been dubbed “superposition.”
Physicist Robert Schoelkopf of Yale University said that when particles can be in two states simultaneously, it offers the potential for unlimited calculations.
“If I can take a quantum computer and I can have that register being in this superposition of all possible states at once, it’s like representing all the possible numbers I might want to multiply together,” he said. “Then I do the computation. In a way, it sort of explores all these possibilities at once.”
The result would be a massive parallel computer.
While the worldwide race heats up in an effort to trap the quantum effects for practical use, a polite scientific competition is playing out in the U.S., as well. Among the chief rivals are Schoelkopf and Wineland.
Schoelkopf runs microwave photons through what look outwardly like standard circuits.
“Basically, our approach in the end is just sort of a custom integrated [radio frequency] circuit,” he said.
There are no laser beams or exotic particles to be generated.
“It’s something you can visualize scaling to a larger number of pieces,” he said.
Wineland’s Colorado setup is more exotic. He suspends beryllium ions in electric fields generated by microscopic electrodes housed inside metallic, thermos-like vacuum tubes. Then he sends ultraviolet lasers through quartz windows to read the ions’ quantum energy states. Each ion scatters light in the 1 state and stays dark in the 0 state, allowing Wineland’s team to conduct basic calculations.
Schoelkopf, though, is politely skeptical that Wineland’s work will lead to a functional computer.
“Dave Wineland’s group has really been one of the leaders in doing quantum information processing with ion systems,” Schoelkopf said. “It’s just that it’s not really so clear yet how that will scale into thousands or millions of quantum bits at some stage.”
Wineland takes Schoelkopf’s critique in stride. “Anybody can point fingers and say, ‘Well, there’s this problem and that,’ ” he said. “The fair answer is we all have big problems to solve to make it scalable.”
And Wineland has his own hesitation about Schoelkopf’s efforts, including the difficult conditions required for the photon experiments.
“Theirs are very sophisticated cryogenic chambers to get to the very low temperatures they work at,” he said.
So whose concept is better? The question is crucial, and it means assessing not just the Schoelkopf and Wineland concepts but research at numerous universities and at companies such as IBM. It’s been hard for government agencies to know whether they are backing potential winners.
In June, Raytheon BBN Technologies of Cambridge, Mass., announced that it had received a $2.2 million Intelligence Advanced Research Projects Activity contract to devise software that could give a “realistic performance assessment” of competing technologies.
“One of the main aims of this program [is] to get a better feel for the resources — what’s important, what is not — so that we can then maybe start evaluating different architectures and different implementations,” said Marcus Silva, a theoretical physicist at Raytheon BBN.
On top of the problem of figuring out how a quantum machine would even work, the biggest challenge for researchers has been to reduce the predicted errors. It turns out that a particle’s quantum state — being in two places at once — can collapse, making calculations unpredictable. That problem is precisely a function of the mysteries of quantum science.
If you put information in a standard computer and act on it, “it does it right,” said George, the former NSA encryption guru. “Quantum computing only does it probably right.”
Reducing bit errors means making the quantum effects last longer, which researchers call “coherence time.” Raytheon BBN is working on that problem under a quantum circuit project. The quantum state needs to last enough fractions of a second, however tiny, to conduct a computation.
“Five years ago, we were at the 100s of nanoseconds range,” said Raytheon BBN’s Zachary Dutton, the theoretical physicist who manages the company’s quantum information group. “Now we’re at 100 microseconds. We’re talking about three orders of magnitude [improvement] in five years.”
Dutton said the problem is close to being solved.
That’s good progress, but some observers expect more twists and turns. “For the past 15 years or so, people have been saying [quantum computing is] 25 years away,” George emphasized. Still, even George says it would be wise to prepare a defense against quantum’s code breaking. George’s former colleagues have a term for what’s needed — “quantum immune encryption” — but as yet there is no mathematical solution or practical technology to do it.
“It’s hard to schedule good ideas,” he said. “They come when they come.”