Quantum Untangled: Taking qubits to the edge
The days of oversized quantum machines being confined to purpose-built labs will soon be a thing of the past.
Close your eyes and think of a quantum computer. I’d put money on the fact that most people reading this pictured a enormous, glistening, chandelier-like cooling system tapering downwards. But the reality is much more mundane. From the outside, a photonic quantum computer, for example, is largely indistinguishable from any bog-standard, rack-mounted machine.
The picture becomes even more diverse when you expand it out to the full quantum ecosystem, bringing in networking, sensors and navigation. It is this wider ecosystem that is likely the first to find a commercial advantage. Aside from niche use cases, around financial analysis, fraud detection and cybersecurity, commercial advantage for quantum computers is still a few years away, as I explained in last week's newsletter.
Quantum Brilliance predicts its accelerators could be used in edge devices and for machine learning
The quantum realm is already having a significant impact on the design and development of next generation technology, including measuring the behaviour of atoms in motion for GPS-free navigation, creating microsecond accurate quantum clocks, and building ultra high definition displays with quantum dots - tiny particles that can exhibit quantum behaviour. Quantum Brilliance is using diamonds to put its quantum accelerators in cars, satellites and edge devices. None of these things bear any resemblance to baroque light fittings.
But the spread of quantum hardware comes as a study from Moody’s revealed businesses are “woefully unprepared” for the coming quantum revolution. As quantum companies being to commercialise and integrate their technology into wider systems, the point where they need to be prepared is coming closer all the time.
In the past week a team from Imperial College London announced its GPS-free navigation system, known as a quantum compass, would be deployed on the XV Patrick Blacket research vessel during an upcoming mission.
This is a form of quantum accelerometer that uses ultra-cold atoms to make accurate measurements. As the cold atoms move through the sensor an optical ruler is formed by using a series of laser pulses that allows the acceleration of the atoms to be measured precisely. This solves some of the "measurement drift" experienced by existing GPS-free navigation systems currently in use.
The team hope that turning to the quantum realm will allow for a much greater degree of accuracy in scenarios where GPS satellites might not be accessible, or where the accuracy or reliability of the navigation tools is paramount. Professor Ed Hinds, from the Centre for Cold Matter at Imperial, said: “I think it’s tremendously exciting that this quantum technology is now moving out of the basic science lab and being applied to problems in the wider world, all from the fantastic sensitivity and reliability that you can only get from these quantum systems.”
The ‘Nvidia of quantum’
This is a good example of an edge computing use for quantum technology. Where you have sensors, you could also make use of computers and accelerators. That is what Australian-German start-up Quantum Brilliance is hoping to achieve with its synthetic diamond-based quantum processors. It is based on defects in the crystal structure of diamond which can be used as qubits. These defects, known as nitrogen-vacancy centers, can be manipulated and read out at room temperature.
The company hopes to become the "Nvidia of quantum” (this may be news to Nvidia itself, which is keen to claim that moniker) by producing quantum processors at scale for a wide variety of use cases. I spoke to CFO George Robinson and the company’s head of software and applications, Florian Preis, at Commercialising Quantum earlier this month. Tracking me down at the coffee machine, Robinson enthused at the potential of a technology that doesn’t require a large data centre operation or extreme cooling in order to provide value.
The goal is to create an “ecosystem” rather than competing with the big players like IBM or Google. This includes building an open-source SDK called Crystal that can be written to in Python or C++ and allows for integration with other platforms.
Quantum Brilliance says it is working on scaling its machines to provide value at the edge
“It's basically off the shelf, high frequency and microwave frequency emitters,” Preis said of the benefit of using diamonds. “It's relatively simple photonics in order to initialise and extract information from the qubits and that's it. This allows us to miniaturise and make the system very robust and therefore going to the edge.” This runs counter to the high performance computing approach most quantum computing companies are taking.
During a follow up ahead of publishing this newsletter, Quantum Brilliance head of product, David Ryan, explained that the roadmap is focused on achieving “quantum utility”, which is the point where it is able to solve real world problems. He said it was “especially interesting in the context of the kind of hybrid and parallelised solutions one can build when they break free of the chains of ‘giant supercooled facility somewhere in North America’.”
He told me that it means they can break free from the endless “science fiction quest for a million perfect qubits”. Which is the long-term goal of almost all the superconducting and trapped-ion quantum computing companies. “I am more interested in the problems that our users and future customers are experiencing, and how our unique offering of a small form-factor, room-temperature quantum accelerator capable of hybrid and parallelised operation can create commercial value for them,” Ryan explained.
Limitations and the future
There are limitations, as is the case with all forms of quantum computing. This includes solving the error problem and tackling scale. Robinson told me the problem was getting the message out that quantum isn’t just about super-cooled giant machines. “rom the commercial perspective the other challenges are that there's already a very loud narrative in the market that cloud is the solution being spoken by a number of companies,” he said. “The diamond quantum model which is our model is the complete opposite. But we believe both can co-exist and they will.”
Like all types of quantum computing there will also be certain types of processing and algorithm that work better than others. For diamonds it will be machine learning. “I see in particular quantum machine learning as one of the opportunities there because you can go to the edge, for example, you can couple it very tightly with existing GPUs. So it's that edge approach and you can reduce the latency between different machines,” enthused Robinson.
William Clements of ORCA says photonic QPUs work well at training machine learning algorithms
This is also something other photonic companies are hoping for, at least as a way to achieve early commercialisation. British photonics start-up Orca has been working with Nvidia on a QPU/GPU hybrid approach that incorporates CUDA Quantum, Nvidia's open-source library for linking the two types of computing hardware together.
I had a far reaching conversation with William Clements, head of machine learning at Orca Computing that covered everything from working with GPU’s to the potential future of quantum computing if all the errors are sorted out. He told me: “It is good to think of the QPU as something that helps the GPU in the learning process”. Basically it takes over part of the learning - in this case it supplies the ‘placement’ details for pixels in a generated image.
Down the road, Clements predicts the photonic hardware will eventually be small enough and scalable enough to work as an edge device. He said one of the main advantages of a photonic system is that it is room temperature. This means it “can be rack mounted in a data centre which will reduce the latency between the different processing types.” It also means it can be more widely deployed than a superconducting qubit processor that requires temperatures close to absolute zero.
Whether its creating a local replacement for GPS by measuring atom movements, using the impurities in diamonds to create a quantum processor small enough to put into a car, or building rack-mountable photonic QPUs that can be deployed in a data centre alongside an army of GPUs, the married of edge and quantum is only just beginning.
Partner content
How do we restore trust in the public sector? - The New Statesman
Defining a Kodak culture for the future - The New Statesman
Brands must seek digital fashion solutions - Tech Monitor
Green bonds and the urban energy transition - Capital Monitor