By Tushna Commissariat in New Orleans, Louisiana, US
“We should have a 20-qubit chip ready very soon… in the next two months most likely,” says Google’s John Martinis to me as we lean against a wall in a relatively quiet corner of the sprawling convention centre in New Orleans. Martinis, a tall man with a mop of silver hair and an Alan Alda-esque manner, is very busy and I’m lucky to have caught him. Earlier that day, he gave one of the APS March Meeting‘s most popular talks on “quantum supremacy: checking a quantum computer with a classical supercomputer”. Martinis, whose team is based at the University of California Santa Barbara, spoke about how they are working towards developing a “scientifically and commercially useful quantum computer” made up of 50 qubits – a 7 by 7 array of superconducting qubits (each of which is coupled to its nearest neighbour) that can be programmed with a one or two-qubit gate – which has an error rate of about 0.1% and can actually do quantum computations.
This is where the concept of “quantum supremacy” – being able to conclusively demonstrate the ability to perform a calculation that a classical computer cannot – comes into play. To check the validity of such a computation would require today’s best supercomputers and, according to Martinis, the most they could compete with is a 50-qubit quantum computer – any more qubits and a supercomputer won’t be enough. Despite that, Martinis tells me that as we enter “the era of a lot of qubits”, using quantum supremacy as a benchmarking tool to see just how well your system works is crucial. For their current 9-qubit simulator, “Supremacy gave us very valuable data and we have extremely precise algorithms for it,” he says.
Apart from that, the Google team is also working hard to predict and fix any possible hardware problems that may arise as the number of qubits is pushed higher, each of which needs to be “very good and coherent”. This includes some specific system engineering requirements. To design better quantum computing processors, Martinis’ team hired Canadian company Anyon Systems. By good fortune, the company’s CEO Alireza Najafi-Yazdi was also at the March meeting and even gave a press talk. Najafi-Yazdi, an engineer by training, co-founded the start-up in 2014 to develop software and simulation systems that help design and optimize quantum electronics and other nanoscale devices where quantum effects come into play.
Anyon Systems specializes in developing “massively parallel” computational tools and simulations, using supercomputers to help design the hardware for future quantum computers. The company is currently collaborating with Martinis’ team to “use Anyon Systems simulation software to predict the cross talk between different components of a 6-qubit quantum processor” explains Najafi-Yazdi, adding that “a comparison between experimental measurements and numerical results shows excellent agreements, proving the promise of massively parallel toolkits in engineering novel quantum processors”. He adds that certain simulations have already revealed potential problems with the superconducting chip that could be addressed during the simulation, rather than spending months building the device only to find it does not function as expected.
Before Martinis and I part ways after our quick chat – he mentions that so much gets discussed at these sessions that his entire team gets together each evening to discuss ideas over a few beers – he says that the team is really pushing to get to a 20-qubit chip very soon. “It all depends how our final fixes go. Google’s stretch goal is for our 50-qubit computer to be ready by the end of this year. It may happen, but I’m more focused on perfecting the 20-qubit chip right now.” In any case, Martinis is confident that the 50-qubit isn’t that far away either and its just a matter of time before Google has a functioning and relatively error-free quantum computer in its sights. Exciting times indeed.