Having performed a quantum calculation that seems well beyond the capability of classical computers, Chinese researchers – with a little Dutch help – appear to have ushered in the “age of quantum primacy.”
In the fall of 2019, Google researchers claimed the world’s first demonstration of quantum supremacy, ie having a quantum device solve a problem that no classical computer can solve in any feasible amount of time (it’s also referred to as quantum primacy or quantum advantage). Their 53-qubit Sycamore quantum computer based on superconducting circuits took 200 seconds to perform a calculation that the researchers estimated would take a top-of-the-line supercomputer 10,000 days to complete.
While experts agree that Google’s result is a milestone in quantum computing, not all of them feel it qualifies as the world’s first demonstration of quantum supremacy. IBM, for example, claims that its Summit supercomputer can actually solve the problem in 2.5 days, and it’s conceivable that a more efficient algorithm will be found to close the gap even more.
As the difference in computation time grows, however, it gets increasingly unlikely that classical computers can compete with their quantum counterparts, for certain types of calculations that is. In that respect, the world’s second attempt at quantum supremacy was much more convincing. Last year, a team led by Jian-Wei Pan of the University of Science and Technology of China (USTC) demonstrated a quantum advantage of about 1014 using a photon-based quantum computer.
Recently, with theoretical contributions from University of Twente researcher Jelmer Renema, the Chinese scientists shattered their own record by performing a similar calculation 1024 faster than a supercomputer. This result makes “classical spoofing of these demonstrations increasingly unlikely and thus establishes more firmly that we’re in an age of quantum primacy for computing,” according to Barry Sanders, a quantum physicist from the University of Calgary, who wasn’t involved in the research.
Twenty years ago, it didn’t look like photons were in the running to demonstrate quantum supremacy: building a ‘universal’ photonics-based quantum computer would require millions of lasers and other optical components. Then, in 2011, theoretical physicists Scott Aaronson and Alex Arkhipov introduced the concept of boson sampling as a technique to demonstrate a quantum advantage with much less resources. The Chinese researchers used a particular variant of boson sampling, called Gaussian boson sampling, which is the most practical to implement experimentally (though still very complicated).
A boson sampling setup is analogous to a Galton board, a vertical board with alternating rows of pegs placed above collection slots. Beads are dropped from the top, and as they find their way down bouncing left or right at every peg they encounter, a bell curve is formed at the bottom. This distribution is straightforward to calculate.
In boson sampling, the beads are photons (photons are part of the boson family of elementary particles) and the pegs are mirrors, beam splitters and other optical components. As photons are ‘injected’ into the circuit, they can interfere with each other as they move through and eventually ‘land’ in one of many output ‘slots.’ Because quantum interactions are involved, calculating the distribution from a given input configuration is much harder. The USTC setup had 1043 possible outcomes.
The photonic quantum computer is good at solving boson sampling because it simulates the set of quantum processes directly, allowing photons to interfere and sampling the resulting distribution. Essentially, the boson sampling setup is a quantum computer that solves itself by ‘being’ the distribution of photons that needed to be calculated.
Now, you might call it cheating to pick a problem intrinsically quantum in nature to demonstrate the superiority of quantum computing. But simulation of quantum processes such as molecular interactions and chemical reactions is very useful. In fact, since these processes can only be simulated classically on a limited scale, (photonic) quantum computing is expected to be a real boon for drug development, materials science and other fields rooted in chemistry. Additionally, certain problems that aren’t inherently quantum in nature, neural networking for example, can be formulated as if they were.
It has been suggested that the boson sampling problem has applications in quantum chemistry. In that respect, it may be more than a convenient vehicle to demonstrate quantum supremacy. To solve practical problems, however, the photonic circuit needs to be programmable, and the USTC setup isn’t. Several startups are currently deploying integrated photonics to build programmable and scalable quantum computers-on-chip. Enschede-based Quix, co-led by Renema, is one of them.