What has happened?
Researchers at Google claim they have built a quantum computer that has solved a random number problem – in just three minutes 20 seconds – which even the most powerful conventional supercomputer would take thousands of years to crack. In other words, a quantum computer has solved a problem that cannot realistically be solved by a traditional computer – the broadly accepted definition of a long-anticipated milestone in computer science known as “quantum supremacy”. Google’s findings were published last week by the science journal Nature and have been widely hailed as a major breakthrough along the road towards realising the immense promise of quantum computing.
What is quantum computing?
It’s a relatively new and still-maturing branch of computer science that draws on the mind-bending weirdness of quantum mechanics – ie, that branch of theoretical physics that describes nature at the smallest scale of subatomic particles. Traditional computers, even the most powerful supercomputers now existing, are built around pieces of data called bits, which represent either ones or zeroes. Ultimately, these computers depend on the flow of electricity: whether a switch is on or off inside a circuit is what generates the ones and zeroes. And readily comprehensible rules for turning the switches on and off can be used to solve mathematical problems by carrying out tasks “in serial”, or one by one.
And a quantum computer?
When you apply quantum mechanics to computing, “things get exciting, not to mention a bit bonkers”, as Ed Conway puts it in The Times. Quantum computers aren’t built around the flow of electricity and on/off switches. Instead, qubits (quantum bits; the unit of information in quantum computing) rely on the properties of electrons, photons and other subatomic particles – and in particular a state called “superpositioning”. What this means is that a particle can have two different amplitudes at the same time, which is roughly analogous to being a one or a zero or (crucially) both simultaneously. Particles are also subject to “entanglement”: a change in one particle instantaneously changes another. In practical terms, this means that quantum computers can carry out an almost limitless number of calculations at the same time, thus increasing computing power exponentially.
What did the Google test actually do?
So far, quantum computing has mostly been a theoretical exercise and Google’s test was squarely in that realm. To demonstrate “supremacy”, they created a machine called Sycamore using 53 qubits – which between them represent nearly ten million billion superimposed states – and set it the task of proving that a series of figures produced by a random number generator were truly random. Pulling it off in 200 seconds was an astonishing feat, though not all scientists agree that it demonstrates “quantum supremacy” – in part because the definition of what “supremacy” means in this context is not a strictly testable standard that is universally shared by those in the field.
Who’s not convinced?
Researchers at IBM say the supremacy claim is wrong. In the Google paper, it is claimed that IBM’s Summit supercomputer would take around 10,000 years to solve the problem in question. But IBM says it could actually take just two and a half days. That’s a lot longer than three minutes, 20 seconds, of course. But such a timescale does mean that the problem solved by Google’s quantum computing is not actually beyond the realistic reach of existing supercomputers. No “supremacy” yet, in other words. Google’s machine is more of a special-purpose device designed to solve a contrived problem without a specific practical application. Yet this still represents a great leap forwards.
What is the use of it?
Researchers in the field believe there are three broad types of problem that quantum computers are especially well suited to handling. The first involves analysing the natural world and modelling the behaviour of molecules with previously impossible precision – a benefit with the potential to revolutionise the chemicals industry and materials science. The second field is machine learning; quantum machines are well suited to handling certain types of probabilistic algorithms (namely those that rely on an element of chance to alight on the best result). A third area is complex optimisation problems that have too many variables for conventional computers; this could be crucial in the design of new drugs and materials, for example.
And this will change the world?
It might. In terms of other real-world applications, the kinds of sectors that are predicted to benefit – or get disrupted – include cryptography, financial services, weather forecasting, battery production and renewable energy. One key hope is to find a better way to use nitrogen from the atmosphere to make fertiliser – a process that currently accounts for 2% of the world’s energy use each year. It’s important to note that the whole field is still very much in its infancy. Google, for example, has yet to start work on the most difficult challenge in quantum computing: the process of error correction needed to clean up the calculations produced by the inherently unstable systems involved. And the scale is still small: the Nature journal reckons that Sycamore’s 53 units is “a tiny fraction of the one million qubits that would be needed for a general purpose machine”. Yet the breakthrough is massive, because it strongly suggests that the “promise of quantum technology can be realised in practice as well as theory”, reckons The Economist – and will suck money and attention into the sector. “A great deal of engineering work remains before quantum computers can be used for real-world tasks. But that day has suddenly got closer.”