It has been a bit of a bumpy start, but a new era of computing seems to be here. Researchers at Google claim their quantum computer has solved a problem that would take even the very best conventional machine thousands of years to crack.
The milestone, known as quantum supremacy, represents a long-sought stride towards realising the immense promise of quantum computers, devices that exploit the strange properties of quantum physics to speed up certain calculations.
Want to find out more about quantum computing? Come to our Quantum World Instant Expert event in Bristol
“This is a wonderful achievement. The engineering here is just phenomenal,” says Peter Knight, a physicist at Imperial College London. “It shows that quantum computing is really hard but not impossible. It is a stepping stone toward a big dream.”
The paper outlining the work, published in the journal Nature this morning, comes a month after a draft was unintentionally posted on a NASA server. It demonstrates that a quantum processor consisting of 54 superconducting quantum bits, or qubits, was able to perform a random sampling calculation – essentially verifying that a set of numbers is randomly distributed – exponentially faster than any standard computer.
Google’s Sycamore device did it in just 3 minutes and 20 seconds, although one of the qubits had to be turned off as it wasn’t working properly.
The latest version of the paper appears to contain no significant changes from the leaked one. For instance, it stands by the claims that the calculation would have taken IBM’s Summit, the world’s most powerful supercomputer, some 10,000 years.
IBM has already pushed back on that claim, insisting that with some clever classical programming, its machine can solve the problem in 2.5 days. Indeed, IBM, which has its own 53-qubit quantum computer, prefers a higher threshold for quantum supremacy, which explains its argument that Google has not yet reached the milestone.
But such caveats should not detract too much from Google’s achievement, says Knight: “There are always going to be clever ways to tweak classical algorithms, but until we’ve had a chance to digest the methodology IBM are proposing, it is hard to judge.”
We will see more of this sort of back and forth when it comes to claims of quantum supremacy, says Knight. “That’s what we expect in any scientific endeavour,” he says. “We always challenge things, and this is such an important development that, of course, we have to poke all of the assumptions.”
Even if you accept IBM’s claims at face value, Google’s quantum computer is still a big step forward, says Ciarán Gilligan-Lee at University College London.
“IBM is claiming that, even when running the world’s largest computer for two and half days, and running petabytes of memory, they can simulate what the quantum chip does in 200 seconds. When you put it into context, it is still a pretty impressive achievement.”
It doesn’t mean quantum computers are ready to tackle real-world problems though – that remains decades away. Instead, it is a proof of concept. “But it is the first baby step on a long road to getting useful quantum computers,” says Gilligan-Lee.
He is looking forward to the next milestone: proof that we have sufficient control over the qubits that we can overcome the small errors they accumulate during calculations.
“We are now in a phase we call noisy intermediate-scale quantum computing, or NISQ,” he says. “To get beyond that, we need to start doing error correction. The nice thing is that we can see from this paper that the architecture of the Google chip is already optimised for that.”