Has quantum computing been cracked?

In recent days there has been a surge in interest in quantum computing - computers that use quantum particles as the equivalent of bits. Out of the blue, I've received several invitations to talk to people about quantum computing as a result of my my book, imaginatively named Quantum Computing, which provides an introduction to the field. I suspect this upsurge is because of the recent announcement that the BBC dramatically headlined Quantum breakthrough could revolutionise computing. 

This is a topic that has suffered from considerable hype in the past - so is this breakthrough (which there certainly has been) transformative or an incremental step towards what is still a fairly distant proposition?

The reason quantum computers are of huge interest is that for certain applications they can, in principle, carry out calculations that would take conventional computers the lifetime of the universe to churn through. The reason that they can do this is that instead of using bits that can store values of 0 and 1, the quantum computer uses qubits - each a quantum particle which can be in a superposition of states - partly 0 and partly 1 simultaneously, with the 'partly' effectively capable of representing an infinitely long real value. The way that qubits link together means that what would usually require sequential processes in a conventional computer can be undertaken simultaneously.

However, there also plenty of problems with making quantum computers work. You need to be able to isolate quantum particles from their environment, or the states of the qubits will be lost, while still being able to interact with them. This is not trivial and as yet it has limited quantum computers to orders of magnitude around 100 qubits. You also need to undertake error correction, because the process is inherently prone to errors, which means it takes considerably more qubits to undertake a calculation that might otherwise be thought. What's more, you need to have both a suitable algorithm, specifically devised for a quantum computer, and the ability to get information in and out of the computer, when the typical answer may well just be 0 or 1.

It's important to emphasise that quantum computers are not desktop devices - they may well always require a specially controlled environment, working as shared cloud devices - and they are not general purpose computers, with relatively limited numbers of potentially very powerful algorithms. The first two examples  produced were an algorithm that effectively makes it easier to crack the encryption used for internet payments (a trifle worrying), and (the reason Google, for example, is very interested) a search algorithm that makes it possible to find something with the square root of the number of searches required by a conventional computer. To emphasise how much the development of this hardware is a slow process, these algorithms were both developed in the mid-1990s, long before anything was available to run them on.

The breakthrough that is making the news involves one class of quantum computers - those where the qubits are based on ions (atoms that have gained or lost electrons to become electrically charged). Other quantum computers use photons, for example, but ions have the advantage of being relatively easy to keep in place due to their electrical charge. A chip to confine and interact with ions requires a lot more space that dealing with the equivalent number of conventional bits. A standard-sizes chip can only handle around 100 qubits, where an effective quantum computer might require a few millions (still vastly smaller than the billions of bits in a conventional computer processor). The breakthrough involves being able to transfer ions from one chip to another with a very low loss rate and without measurably impacting the 'phase coherence' of the qubit - in simple terms, the qubit keeps the values its holding.

This is an impressive piece of work. It makes it possible in principle to have a quantum computer with many chips that interact with each other, enabling it to support the kind of number of qubits that would make it a truly effective resource. However, it's worth emphasising that there are still plenty of other issues to be dealt with, and that while this is an effective demonstration, it's still a way from being applicable on any scale. Realistically it could be another 5 to 10 years before there is a real product where large scale, useful quantum algorithms can be deployed. An important step, then, but definitely incremental rather than a revolution.

If you'd like to read more about the technology, the paper is here and is freely downloadable. (Surely it's time the BBC started providing links to papers?)

Feature by Brian Clegg - See all of Brian's online articles or subscribe to a digest free here