Google, NASA put big money on D-Wave's quantum computer
Promising results, but critics say technology might not live up to theoretical benefits
Google. NASA. Lockheed Martin. Los Alamos National Laboratory. Big names in the worlds of big brains and cutting-edge technology are investing millions of dollars in the quantum computing technology of the Burnaby, B.C.-based company D-Wave. They say they're starting to see promising results, despite criticism from some quantum physicists that the technology might never live up to its promise.
Quantum computing is a new form of computing based on quantum mechanics, the strange physics that affects very, very small particles such as atoms. In theory, it has the potential to revolutionize artificial intelligence, space travel and other fields by solving problems conventional computers either can't solve or can only solve very slowly.
But even Google acknowledges that D-Wave's limited form of quantum computing hasn't yet proven capable of doing anything conventional computers can't. And critics say companies that invest in it are making a "wild bet."
So, why are so many brainy organizations doubling down?
Repeat customers
Late last year, D-Wave sold a new system to Los Alamos National Laboratory, best known for developing the first nuclear weapons during the Second World War.
It also renewed multi-year contracts that include regular upgrades to the latest model of its quantum chip, with both military technology company Lockheed Martin and Google, in partnership with the Universities Space Research Association and NASA.
When asked why, Harmut Neven, Google's director of engineering, noted that D-Wave is currently the only company in the world that sells quantum computers commercially "and I just wanted to make sure for the forseeable future, we have access to the latest and greatest chip."
All of these organizations hope to harness the unique advantages of quantum computing over conventional computing.
How it works — in theory
Quantum computers manage information as qubits instead of bits – the 1s and 0s used by conventional computers to represent data.
While the power of conventional computers is proportional to the number of bits, the power of quantum computers grows exponentially with the number of qubits.
That's because qubits can represent both "1" and "0" at the same time — a phenomenon called "superposition" — allowing them to perform multiple calculations simultaneously.
So far, most quantum computers being built by researchers around the world have only a very small number of qubits, but they're constantly getting larger and more powerful.
Soon, says Raymond Laflamme, executive director of the Institute for Quantum Computing at the University of Waterloo, "you'll be able to solve problems that classical computers just cannot do."
Theory suggests that quantum computers should be particularly good at solving problems that involve searches of huge data sets, as they can search multiple states at the same time.
"This is the place that the Googles of the earth are interested in," Laflamme says.
On the website for NASA's Quantum Artificial Intelligence Laboratory (QuAIL), the space agency says it is hoping to use quantum computing to help sift through data from the Kepler telescope and discover new exoplanets.
"Quantum computers may theoretically be able to solve certain problems in a few days that would take millions of years on a classical computer," QuAIL's website says.
'Bread and butter' of engineering
That's particularly the case for "optimization problems," where you try to find the best possible combination among a huge number of parameters.
Neven says that's "bread and butter for pretty much any engineer."
"Say you want to have a more efficient car that has less air resistance or you want to build a better car battery, schedule your airline flights better or offer more attractive financial options. Optimization is always the key task you have to solve."
That also applies to machine learning, a type of artificial intelligence that involves computers teaching themselves, which is what Neven says he and Google are most interested in.
Solving optimization problems becomes exponentially more difficult for conventional computers as the number of parameters grows. But the difficulty should only increase linearly, rather than exponentially, for quantum computers.
That's the theory, but does it actually work?
Preliminary results published by Google say yes.
100 million times faster
In December, Neven reported that D-Wave's system can be more than 100 million times faster than a conventional computer when using a similar method to solve an optimization problem with 1,000 variables that could each be in one of two possible states.
D-Wave's quantum computer is a limited version called a quantum annealer.
It's not a "universal quantum computer" that can solve all problems — something that researchers around the world are trying to build. But a quantum annealer can theoretically solve optimization problems faster than a classical computer.
Quantum annealing involves converting an optimization problem into a 3D landscape of hills and valleys, where the deepest valley is the best solution.
D-Wave's Colin Williams says classical computers solve the problem by hopping from point to point.
A quantum physics phenomenon called quantum tunnelling allows D-Wave's computer to do more than that.
"Instead of hopping on surface landscape, you can actually tunnel through the hills," Williams says. That allows it to find the solution more quickly and easily.
Neven says the new results prove that D-Wave's chip really does make use of quantum physics to outperform classical chips.
"That was very exciting," he told CBC News.
A wild bet?
They also counter some previous criticism of D-Wave from quantum physicists.
But not all. The University of Waterloo's Laflamme says that while it's been mathematically shown that universal quantum computers will be able to solve problems that conventional computers can't, the same can't be said of quantum annealers:
"With the D-Wave, the theory is not as much developed," he said.
On the other hand, D-Wave is promising companies like Google a potential entry point into the brave new world of quantum computing,
"Maybe they're thinking, 'Let's make a wild bet and see if this can happen,'" Laflamme said.
Neven acknowledged that for now, there are conventional computers and algorithms that perform better than D-Wave's machine at solving optimization problems, "but that is for today's generation of machines."
Working with D-Wave's chip allows Google researchers to pinpoint what improvements need to be made in order to outperform classical supercomputers — something that would be impossible to do using just a classical computer, he said.
NASA researchers, like Google's, are plowing ahead. They're now converting problems in areas such as planetary rover exploration into a form that can be processed by D-Wave's chip.
D-Wave's Williams says that's easily done. "You don't need to know any physics at all," he said.
Early days
But Rupak Biswas, deputy director of exploration technology at NASA Ames Research Center, says working with D-Wave's computer is kind of like working with conventional computers when they were first invented.
"This machine does not have compilers or programming languages," he said. "By basically turning these virtual knobs and setting some voltages and biases … you run the system, and then the system comes back and gives you a stream of 0s and 1s, and you try to figure out what the answer is."
A quirk of quantum computers is that they give you a different answer every time.
"And so the trick is therefore, how do you know which is the right answer? What any quantum system will do is try to solve the same problem over and over again to see what is the most likely solution, and it may still be wrong," said Biswas.
Neven acknowledges that researchers are still figuring out the theory and don't fully understand why quantum annealing works but says he's increasingly confident that the technology will soon outperform classical computers, achieving something called "quantum supremacy."
"I don't have mathematical proof, but I have pretty strong experimental results," he said. "So, I'm pretty optimistic at this point."
ABOUT THE AUTHOR
