The debate over what kinds of problems can be solved quantum computationally that couldn't even begin to be solved by a conventional supercomputer has been reignited by recent trials using two of the world's most powerful quantum computers to date. The country holding the flag in this regard is China.
Theoretically, a quantum computer of sufficient complexity - enough quantum bits or "qubits", for example - could develop a "quantum advantage" that would allow it to solve problems that a classical computer would never be able to solve. In theory, a 300-qubit quantum computer could handle more computations at one time than the atoms in the universe.
In 2019, Google claimed that its 53-qubit Sycamore processor demonstrated such "quantum priority" and completed a calculation in 10.000 seconds, which the company estimated would take 200 years to power Summit, the world's most powerful supercomputer at the time. But IBM researchers later questioned the quantum advantage claim, claiming that Summit could solve this problem in 2,5 days with better classical methods.
The state of the art right now is that no experiment has shown a quantum advantage for real-world tasks (Chao-Yang Lu, China University of Science and Technology).
Now, Chinese researchers have put two different quantum computers to the test on tasks that Sycamore claimed were more difficult than they had to tackle, and the results were faster. They note that their research shows "a seamless quantum computing advantage."
In one experiment, the researchers used Zuchongzi, a system that uses 56 superconducting qubits, to solve a problem with random samples from a set of possibilities. They discovered that Zuchongzi had completed a sample assignment that would have taken him at least 1.2 years to complete Summit in 8.2 hours. Additionally, they pointed out that this sampling process requires tens to hundreds of times more computation than what Google used to gain a quantum advantage for Sycamore.
In a different experiment, the researchers put the photonic quantum computer Jiuzhang 2.0 to the test by evaluating patches of random data using a technique called Gaussian boson sampling. The problem of Jiuzhang 2.0 using 113 detected photons is about 10 times better than traditional supercomputers.24 They calculated that they could solve it twice as fast.
The Gaussian boson sampling problem on which Jiuzhang 2.0 was tested may have many practical uses, such as determining which pairs of molecules best match each other, although the sample work used in studies using Zuchongzi has no known practical utility. As a result, this work may have quantum chemistry implications in replicating fundamental molecules and chemical reactions, according to physicist Chao-Yang Lu, who is both a co-author of the studies and a professor at the China University of Science and Technology in Hefei.
According to Lu, these new tests are "robust and fundamental milestones for creating more complex quantum computers." However, it does offer a warning about the growing hype surrounding quantum computing.
According to Lu, the number of computational problems that can truly benefit from quantum computing is still extremely small.
“The cutting-edge technology right now is that no experiment has shown a quantum advantage for real-world tasks. The world only needs five quantum computers, so we shouldn't be overly pessimistic or narrow-minded, but we do need it. distinguishing between optimism and exaggeration.”