The first computer program was written by Ada Lovelace in 1843 for Charles Babbage's marvellously ambitious (and tragically hypothetical) Analytical Engine. Intended to showcase the computational power of this whirring, shunting assembly of cogs and flywheels, Lovelace’s code was designed to calculate a sequence of Bernoulli numbers. Though the Analytical Engine was never completed in either Lovelace or Babbage’s lifetimes, computer science students still use Bernoulli numbers to flex their coding muscles.
Most complex machines are now equipped with a silicon chip and a coding language governing the parameters of its mission, whether they’re inside the washing machines scrubbing your clothes clean in a hundred different modes at 30°C (or higher than that, if you’re truly unhinged) or the screaming selection of children’s toys designed to delight and irritate households in equal measure. Quantum computers are no exception to this trend. Being equipped with qubits instead of bytes is all well and good, but you still need a way of telling the machine how to harness those nifty, slippery particles and define their calculations. It should be of little surprise to loyal readers, then, that more than a few companies in the quantum ecosystem are busily producing algorithms, operating systems and abstraction layers that tell the qubits what and how to perform set tasks, allowing the end user to write in Python, C++ or Q#.
To allow these languages to realise their full potential, major players in the quantum space have devised open source frameworks that include all the technology designed to run code on quantum hardware — think Qiskit from IBM, or Cirq by Google. But for a quantum computer to achieve advantage — or ‘utility’, as IBM is now calling it — the software layer needs to be even more abstract and easier to understand.
One company focusing on complete abstraction is Multiverse Computing. It offers a SaaS product called Singularity that, to the end user, appears as nothing more than another menu item in Microsoft Excel. That’s because, in the back end, Multiverse works with its clients to understand their data requirements to a point where they can be boiled down into the tailor-made, quantum equivalent of the flick of a switch.
None of this work necessarily requires a quantum computer, explains Esperanza Cuenca-Gómez, Multiverse’s head of strategy. “We use tensor networks,” says Cuenca-Gómez. “Quantum-inspired techniques use the principle of quantum mechanics and then those algorithms are deployed on high-performance classical hardware.”
Real world results are already being seen and acted on from today’s quantum systems, she adds, even though the software layer isn’t necessarily being deployed on quantum hardware right now. That said, Cuenca-Gómez claims, it’s perfectly capable of being run on tomorrow’s quantum computers as and when they’re ready.
Full article available here.