A quantum computing start-up is developing its own operating system to run quantum machines, akin to Microsoft Windows or macOS on a standard computer.
The hardware for quantum computers has progressed rapidly. Five years ago, few of these machines contained more than 20 quantum bits, or qubits – the building blocks of quantum computers – whereas today devices with more than 100 qubits are increasingly common. But in order to actually make calculations, those qubits require quantum software.
The “quantum stack” – the layers of programs and systems that make it possible to use a quantum computer – will be key for making these devices useful in the near term, says at Rice University in Texas, who isn’t affiliated with the new research. “It’s fundamentally a very different stack. It cannot be copied at all [from conventional computing].”
Advertisement
, a start-up based in Singapore and Ireland, is working on developing the layers of that stack, from fine-tuning the electromagnetic pulses that control qubits to developing an operating system, the program that manages a computer’s hardware and all the other software it needs to be functional. at Horizon Quantum presented the company’s work at the in Santa Clara, California, on 11 December.
To program quantum computers, researchers write quantum circuits, sequences of basic logic operations that make up the computation. But to execute these circuits, the programmer needs to determine the role that each qubit should play at various times during the computation. Which should actively perform the operation, and which should be used as memory while the program is running? Setting those roles is a task that conventional computers can perform.
Fitzsimons and his colleagues developed conventional software that can take a quantum circuit and determine the most efficient way to use qubits to run it. So far, they have demonstrated this functionality for several simple programs with a few qubits at a time. “We’ve built a very general optimisation engine. And we are starting to see foundations of an operating system emerge,” says Fitzsimons.
Allocating different roles among qubits is especially important for certain kinds of quantum computers, such as those made from extremely cold atoms, where qubits aren’t connected to each other in a permanent way. Instead, they can be moved around, connected and re-connected at each step. “We have a lot of great flexibility, and there are many options for how to translate a circuit into atom moves,” says at Massachusetts-based quantum computing start-up QuEra, who wasn’t involved with the new research. He says this is where such quantum software can make a big difference.
Fitzsimons and his team have been testing their software with quantum computer simulators and on physical quantum computers that they accessed through the internet. But now they have also acquired their own quantum computing hardware.
Specifically, the researchers will be using qubits made from tiny superconducting circuits, which they hope will be operational by the end of the year. Fitzsimons says that they also want to work with other types of qubits, like ultracold atoms, because their software is built to be of general use and not specialised to just one type of quantum computer.
The team’s testbed involves multiple companies: the hardware layer was built by , some of the hardware control will be taken care of by systems from start-up and Horizon Quantum will use its own software tools to optimise and implement quantum circuits.
Until recently, all these functions would have been handled by one company, says Patel, and therefore been proprietary – and more narrowly focused on one device. Several specialised companies working together may have a better chance of getting new software capabilities of the ground, he says.
