Classical and quantum computers are vying for superiority
Will 2019 be the year when quantum computers show they have the right stuff? Google says so — one of the company’s labs, in Santa Barbara, California, has promised that its state-of-the-art quantum chip will be the first to perform calculations beyond even the best existing supercomputers.
And Google isn’t alone. A number of other companies, big and small, are working steadily towards the same symbolic goal. Venture capitalists have poured money into dozens of quantum-computing start-up companies. Excitement and anticipation are mounting.
In a stark reminder of the power of quantum computing, in May, two theoretical computer scientists solved a 25-year-old conjecture (go.nature.com/2eatyco). They confirmed that quantum computers are — in an admittedly abstract setting — vastly more efficient than classical ones at particularly complex tasks, such as testing whether a set of numbers is random.
Still, such work does not justify the expectations that now surround quantum computing. A recent report by the US National Academies of Sciences, Engineering, and Medicine (penned by leading Google and Microsoft researchers, among others) stressed the technical hurdles that lie in the way of building practically useful quantum computers. Creating such machines will take at least a decade, the report says.
Theoretical physicist Seth Lloyd at the Massachusetts Institute of Technology in Cambridge speaks for many when he says the field is in a period of explosive progress — but that the hype is also getting out of control. “The whole quantum-computing field is just going hogwild right now,” he says.
Is a quantum computer even needed? High-profile work by an 18-year-old computer scientist earlier this year suggests not, at least for one specific task. Ewin Tang effectively taught an old computer a new trick — one that was previously thought to need a quantum system.
She developed an extremely efficient classical algorithm — that is, one that can run on an ordinary computer — for ‘recommendation systems’, such as those that certain websites use to try to guess a consumer’s tastes (E. Tang Preprint at https://arxiv.org/abs/1807.04271; 2018). Her work produced a much faster version than current, relatively sluggish systems. Tang’s algorithm is not necessarily practical to use, so it won’t replace current algorithms unless it is substantially improved — in its current form, it would be useful only with data sets of truly gigantic proportions. But a quantum algorithm that was in development for that same task has now been rendered moot, before it ever had a chance to run on an actual machine.
Last month, Tang, who is now at the University of Washington in Seattle, doubled down. She and two colleagues demolished the quantum advantage of another type of algorithm for certain machine-learning tasks (A. Gilyén et al. Preprint at https://arxiv.org/abs/1811.04909; 2018). A different team at the University of Texas in Austin reached the same conclusion independently (N.-H. Chia et al. Preprint at https://arxiv.org/abs/1811.04852; 2018). Computer scientists responded to the news with memes that, for example, compared Tang to a gladiator slaughtering the hopes and dreams of the quantum community. And it was a bittersweet moment for Tang’s co-author, Seth Lloyd — he wrote the quantum algorithm that was trounced.
Some in the field argue that these uses of classical computing are actually successes for quantum computing, because they show how the quantum way of thinking can have an impact, even before quantum computers exist. Specialists also point to problems for which quantum computers have long been known to have a proven advantage, such as web searches. In other cases — such as factoring large integers into primes or simulating the electronic properties of materials — scientists think that quantum computers are still likely to have an advantage, although this has not yet been demonstrated mathematically.
Quantum computers are a not-yet-existent technology in search of problems to solve. Meanwhile, researchers are seeing how far classical strategies can be taken. Both are valid research avenues. A quantum device remains a laudable goal. But it’s not the only route to the future.
Nature 564, 302 (2018)