Talk to those in the quantum computing arena and you will either be enlightened or confused on any given day. At times the language of quantum can be arcane. There is even considerable debate as to whether there is actually a commercially available quantum computing solution.
Vern Brownell, CEO, D-Wave, is sure there is. D-Wave is the first company to build a market-ready quantum computer after over 15 years of R&D. Today, it has installations with Lockheed Martin, NASA and Google, among others.
“The basics of quantum computing sound simple: leveraging quantum mechanics in a way that allows you to do computations faster and more efficiently than is possible with any other forms of classical computing. But the reality is it’s very hard to build these devices,” he said.
Misconceptions about is viability is a result of the fact there are different types of quantum computing models, he added. “Quantum computing comes in different flavours, which is why there may be so much confusion in the press. Distinctions are im
portant when you are trying to understand the industry. When you pick up a book on quantum computing, it is generally only talking about the gate model, so the assumption is there’s only one. That is absolutely not true.”
The gate or circuit (aka universal) quantum computing model has likely had the most investment over time, Brownell said. “Most research institutions are doing gate models, which is a perfectly viable form of quantum computing. To date most gate models have only been able to deliver 5 qubits (Quantum bits), but you need to reach the 10s or thousands to do anything useful. That will take time.”
IBM is working with the universal quantum computing model for its Quantum Experience. When it was originally launched in May of last year, Quantum Experience was an educational initiative to “get people understanding what universal quantum computing is all about,” said Scott Crowder, CTO, vice president of Quantum Computing, at IBM Systems. “It provided a simple graphical interface to all kinds of gates and operations, and allowed users to actually run instructions on real computers to see results.”
The next step ocurred in March of this year, when IBM announced programmatic access to code writing, in addition to the original drag and drop functionality. “It’s still primarily educational, but it also enables some level of algorithm development so we can build out an ecosystem of developers. Now anyone can connect to IBM’s quantum processor via the IMB Cloud to run algorithms and experiments, work with individual quantum bits, and explore simulations.”
IBM’s also intends to build commercial systems above and beyond the public access to the Quantum Experience, he explained. “We need to get this in the hands of industry and academia, not just for educational purposes but for real use cases. So we are going to be building systems for early access partners this year with the intent of upgrading systems for them over time. Over the next few years we plan to get to a 50 qubit commercially available system. The Quantum Experience in use today is five qubits.”
But in the world of quantum computing not all qubits are equal, he said. Which brings us to the second model: the quantum annealer, which Crowder described as “a fixed function system with very different scaling properties.”
Quantum annealing is the model which D-Wave began developing in the 1990s and is in fact commercially available. “It was a controversial choice for us at the time. It was completely greenfield R&D work when we started,” Brownell said.
He provided the following metaphor to explain how quantum annealing works. Consider a 3D landscape terrain such as the Rockies. Your task is to find the lowest valley in that terrain. “Computationally you would have to go through every single alternative. You won’t be sure of the answer until you traverse the entire landscape. Quantum annealing can very easily find the lowest space using qubits rather than traditional simulations by passing through the hills instead of climbing them and then discovering correlations between the coordinates that lead to deep valleys. When fused with artificial intelligence, that capability can be applied to lots of human scale problems like climate modeling, financial investments, risk calculations, etc.”
Further out on the horizon is a third model – topological quantum computing – which Microsoft is researching, Brownell said. “It’s supposedly the purest and most elegant form of quantum computing. But it requires the discovery of a particle that no scientist has seen yet. It may be decades before useful hardware can be built. The advantage of this approach is that there is less error correction than there is with the gate model and it is much more immune to that noise that is the enemy of quantum computing. It’s very exciting, but super hard to do.”
As the quantum computing market heats up and developers come on board, the rivalry among providers continues to be a friendly one. “Everything is good,” Crowder said. “Google, Microsoft, Intel and startups are in this space so it’s a really exciting time for the industry. We’re all trying to push really hard to create more access for more people more quickly.”
This video done by D-Wave partners at Google offers more detail on the applications scientists are investigating and the technology that is currently installed.
IBM has made the specs for its new Quantum API available on GitHub https://github.com/IBM/qiskit-api-py and is providing simple scripts https://github.com/IBM/qiskit-sdk-py to demonstrate how the API functions.