By Andre Saraiva, UNSW
A few months ago, JianWei Pan from the Chinese Academy of Science announced he had broken Google’s record by creating a 50-photon boson sampling quantum computer that cannot be simulated by even the most powerful classical computer. Now, the numbers are out in a much-anticipated paper published in Science magazine. Using the best algorithms around, China’s almighty supercomputer Sunway TaihuLight would take 2.5 billion years to generate the same samples that the USTC quantum computer generated in just three minutes. That is more than half the age of our planet itself.
Aptly named Jiŭzhāng after the legendary ancient maths book Jiŭzhāng Suànshù (The Nine Chapters in Mathematical Art), this quantum computer provides a Chinese lesson on the power of exponentials. With just 25 two-mode photon channels (or 50 single mode photon states) , the quantum correlations generated by the photonic circuit have outperformed Google’s Sycamore by an astonishing factor of at least 250 000 (measured in terms of the time it would take a modern supercomputer to perform the same task).
Arguably, this number could be much larger – there is some dispute regarding the classical algorithm the Google used to benchmark their supremacy claim. But the lesson is that the exact number does not matter – quantum advantage is exponential and there is no escaping the reality that we are reaching the era of uncomputable, man-made quantum states.
The second key takeaway from this work is just how advanced quantum engineering of photonic devices is at USTC. In order to achieve the level of correlations that Jiŭzhāng did, every single part of the quantum computer has to be pushing the boundaries of technology.
The first challenge is to get the necessary efficiency in generating photon, filtering them to have the same frequency, moving them in a 3D circuit and detecting them. Each of these steps induces some losses. To get 100 individual photons to successfully walk through the maze of interference pathways without any of the bailing out is itself no easy feat.
Then comes the hardest challenge which is to make sure that all photons arrive at the precise same time, with an error inferior to one part in a billion. As described in an interview for Scientific American by one of the researchers in this work, Chao-Yang Lu, this “is the equivalent of 100 horses going 100 kilometers and crossing the finish line with no more than a hair’s width between them”. The researchers developed a phase-locking system to compensate environmental fluctuations and give each photon the correct handicap in order to line them up at the end of the path.
So, what is next?
It is now unclear what is the future of this line of research. Arguably a dead-end, BosonSampling can at best be used as a stepping-stone to the far more daunting challenge of constructing a universal quantum computer with linear optics. Even though the article is titled “Quantum computational advantage using photons”, there are few accounts of useful problems that can take advantage of Jiŭzhāng’s massive Hilbert space dimension (read more about this in our other Analysis article about the meaning of this result).
Potentially, if the USTC group works on packaging, creating a user interface and making their quantum computer available on the cloud, users from around the world could drill down on the capabilities of Jiŭzhāng and eventually find an application for it, other than the academic exercise of proving quantum supremacy.
From this aspect, Google’s supremacy experiment is a far more exciting result. The correlated state created by Sycamore involves universally controllable superconducting qubits, which could potentially be used for all the staple applications of quantum computing such as Shor’s factorisation algorithm and Grover’s search algorithm. In practice, the cripplingly high error rates and lack of quantum error correction in Google’s quantum processor also means that it is merely a stepping-stone towards the promised land of quantum advantage.
Dr. Saraiva has worked for over a decade providing theoretical solutions to problems in silicon spin quantum computation, as well as other quantum technologies. He works on MOS Silicon Quantum Dot research and commercially-oriented projects at the University of New South Wales (UNSW)
December 4, 2020