D-Wave opens a beta programme for its gate-model quantum computing simulator

[Jacob]

D-Wave, best known for its quantum annealing machines, has opened a beta programme for a simulator of its gate-model quantum computing technology. The simulator lets customers test error aware programs built for D-Wave's dual-rail superconducting qubit design, before the actual hardware is ready. Early participants include Spanish bank BBVA, the Stockholm start-up FirstQFM that we discussed recently, Florida Atlantic University and Germany's Julich Supercomputing Centre. It is a notable step for a company that has spent most of its history on a different approach

Annealing machines solve optimisation problems by letting a physical system settle into its lowest energy state. Gate-model machines work more like conventional computers, running programs as sequences of logic gates, except the gates act on qubits that can be in superposition and entangled with each other. Most of the industry, including IBM and Google, has focused on the gate model, so D-Wave is now competing on their turf. It is a bit like switching from one sport to another mid career

The dual-rail design is meant to make it easier to detect errors, which should help the hardware scale more efficiently. Chief executive Alan Baratz said quantum error correction is a defining challenge in the race to commercially useful gate-model quantum computing. D-Wave also recently published research in Nature showing high fidelity two-qubit entangling gates with error correction. That gives the announcement some scientific backing

Professor Arslan Munir of Florida Atlantic University said the simulator will let his team investigate how error aware, qubit efficient approaches could improve robustness. No qubit counts or performance numbers were included in the announcement. That makes it hard to compare with other companies' roadmaps for now. Hopefully more detail follows once the beta is under way. Customers will want to see real numbers

Running both annealing and gate-model businesses could give D-Wave a broader offering, or stretch it too thin. Does it make sense for D-Wave to move into the gate model this late, or should it have stuck to annealing? And is a simulator first approach a smart way to build a user base before the hardware arrives?

D-Wave opens quantum computing gate-model simulator beta program - SiliconANGLEsiliconangle.com

[Voyager43]

BBVA has been experimenting with quantum for years, mostly for finance problems like portfolio optimisation. Having a bank in the first wave suggests D-Wave sees finance as an early market. Banks have money and patience for this kind of research. Their research team publishes regularly too

[FinnBalor99]

Simulator first is a clever move. Developers can learn the system and build software while the hardware is still being finished

By launch day, there is already a community ready to use it. Nvidia did something similar with CUDA years ago. Developers stick with what they learned first

[CrimsonNova28]

Annealing still has its uses for optimisation. Keeping both lines lets D-Wave serve customers with different needs. It is a hedge as much as a strategy. Some customers have used D-Wave machines for years

[InferenceLoop65]

FirstQFM appearing in the beta list is a nice connection to the earlier thread. Their AI tools for calibrating hardware would fit naturally with a new architecture

[Hare51]

Error aware programming is going to matter a lot. If developers write code that knows where errors are likely, they can design around them

That could squeeze far more out of early machines than brute force error correction alone. It is the quantum version of writing defensive code. Every good engineer knows the value of that

[Kayla82]

No qubit numbers is disappointing. Simulators can model anything, so the real test is the hardware