Quantum leap
How Microsoft’s quantum computing led to a new particle
The idea that you can make a computer from subatomic particles has been around for a while now. It was first proposed by Russian mathematician Yuri Manin in 1980, and championed by physicist Richard Feynman the following year. A functioning, large-scale quantum computer always seems to be just over the horizon, but never arrives.
Microsoft wants to accelerate the process, and it has set up a research division that not only aims to solve the problems of producing a quantum computer, but also to address its substantial cooling and programming needs. Its California quantum research center—Station Q—was set up in 2005, but the company is working all around the world, including at the firm’s laboratory in Delft, Netherlands, and at the Niels Bohr Institute, part of the University of Copenhagen.
The advantage of quantum computers is that they could potentially solve problems in an afternoon that would take a classical computer a billion years, but this is only true if the algorithm being processed has been written to take advantage of the architecture. Where a regular CPU uses bits—transistors that can be either on or off—a quantum processor uses qubits (quantum bits) that can be in either of those states, every state in between or both. This can hugely speed up computations, for reasons that are… complicated.
Weird science
Before we get into that, though, what exactly is Microsoft doing? “We’re developing a complete end-to-end solution,” says Microsoft Quantum’s corporate vice president Todd Holmdahl. “We’re working on the physics, the hardware to control the physics and the software to control the computer as well as run all the applications and algorithms. We’re also basing our computer on what we think is the world’s most stable qubit, called a topological qubit, and this has three or four orders of magnitude better fidelity than any of the other qubits out there.” And when will it be here? “Five years.”
This claim only makes sense if you know that there are a few different ways of making a quantum computer. One requires superconductors that operate at near-freezing temperatures. Another uses focused laser beams to trap particles. A third measures the ‘spin’ of electrons. But all of these are vulnerable to quantum decoherence—a process where interference from the outside world sees the quantum superpositions (being in two places or states at once) of the particles collapse. Quantum becomes classical. By being stable, the topological qubit can, if not overcome this hurdle, then at least make the overcoming a little easier.
This approach has required the development of novel materials, including a subatomic particle. “It arose out of the theoretical realization that there was a kind of particle with a property that had never been seen before,” says Professor
Charles Marcus, a principal researcher at Microsoft who is also professor of condensed matter physics at the Niels Bohr Institute. “The crazy thing is, we had to create the particles. The property is called, and I’m sorry it doesn’t have a better name, non-Abelian statistics.” There are two fundamental classes of particle: Fermions, which are what matter is made of, and bosons, which include the force particles. Non-Abelian particles are made in a reduced dimensional space.
These particles are the building blocks of Microsoft’s computer. “They exhibit this property that, when you exchange their places, unlike with fermions and bosons, you can change the state of the wave function so you can do computation,” says Marcus. Unlike qubits produced by other methods, topological qubits are very hard to decohere. “These nonAbelian particles encode information in a non-local, distributed way. It’s a very abstract idea, and I wish there was a simpler way to tell the story, but that little bit of magic is the basis of the Microsoft approach.”
Essentially, the movements of these particles create what are known as ‘braids’ in spacetime, and these can be formed into the processor’s logic gates. How the computation happens, however, is different to the operation of a classical CPU: “It’s engaging in something like interference,” explains Marcus. “The trick of quantum computing is to have the right answer appear as constructive interference, and all the wrong answers appear as destructive interference. They destroy each other, and all the correct answers reinforce each other. You have to set up the problem so that it involves an interference experiment in some sense, and the right answer will be at the location of constructive interference.”
Future perfect
Getting a problem set up is no small feat in itself, but Microsoft has thought of this with Q#, a programming language that’s part of the Quantum Development Kit, available online to tinker with now. Elsewhere, IBM has a functional five-qubit quantum computer, based on superconducting principles, plugged into the internet for users to experiment with, and Google has unveiled Bristlecone, a 72-qubit machine. It’s tech that’s nearly here, for real, and it’s looking for applications.
“An example we have studied is nitrogen fixation,” says Holmdahl. “That’s cutting the bonds of a nitrogen molecule to create artificial fertilizer. Today it takes 3% of the world’s natural gas. But we know that nature does it at normal temperatures and pressures, using an enzyme, so we believe that with a quantum computer we can find that enzyme and produce fertilizers at much lower cost. That’s a big one, and you can imagine applications such as finding a catalyst to sequester carbon from the air too.”
Marcus believes we are only scratching the surface of what quantum computers might be capable of: “When we really have the ability to control a quantum mechanical system, why don’t we just ask it what it’s good for? It should be smart enough to answer that question. It’s going to be a lot smarter than we are.” Ian Evenden
It’s tech that’s nearly here, for real, and it’s looking for applications