Chinese quantum computers: a chip above the rest
A year ago, a Google quantum computer outperformed classic computers for the first time. Now a Chinese team has also succeeded in doing so – using a completely different technique.
A Chinese quantum computer prototype has solved a computational task that would take a classical computer 2.5 billion years. In 200 seconds. This was recently reported by researchers led by Jian-Wei Pan from the University of Science and Technology of China in the scientific journal «Science».
A Google-funded research team had already shown that quantum computers can achieve their goals faster than conventional supercomputers on a special-purpose task back in October 2019. The US team used a network of 53 superconducting crosses for this purpose. In record time, the heavily cooled microchip determined the probability with which the coupled qubits would produce certain sequences of numbers when read out.
The Chinese team chose a different path to «quantum superiority»: using lasers, the physicists fired a flock of individual light particles into a field of beam splitters, mirrors and prisms. At each of the beam splitters, a photon must decide whether to turn at a right angle or continue flying straight ahead. At the edge of the experimental area, 100 sensitive detectors showed which path the light particles had ultimately taken across the laboratory table.
Normal computers struggle greatly with calculating what percentage of detectors will light up and what polarisation the incoming light particles will have in each case. If you add up all possible routes through the course for 50 photons fired simultaneously, you arrive at the astronomically high number 10³⁰ (a number with 30 zeros), according to researchers' estimates. And ultimately, not all outputs are equally likely: the light pulses can be described as waves within the framework of quantum physics and therefore interfere with each other in a complicated way.
To determine how frequent which sequence of the experiment is, classical computers have to solve a tricky system of equations. Its complexity grows exponentially with the number of light particles involved. If, on the other hand, one simply replicates the course, regularly feeds it precisely coordinated laser pulses and counts how often every photodetector lights up, the result can also be determined much more quickly: in this case, nature relieves one of the complicated superimposition processes.
Computer scientists Scott Aaronson and Alex Arkhipov had already described the experiment known as «Gaussian Boson Sampling» in 2011 as a testing ground for a quantum computer based on light particles. Until recently, however, the technical hurdles were considered extremely high, since light particles are often lost and are difficult to synchronise in large numbers.
However, the Chinese researchers now claim to have succeeded: they measured up to 76 photon signals per run of the experiment, the team reported in «Science». This has reached a level of complexity that can no longer really be simulated by supercomputers. Consequently, the scientists were only able to determine whether the «probability distribution» spat out by their quantum computer was correct in a roundabout way: among other things, they used supercomputer simulations for up to 50 measured photons and a comparison with laboratory results in which light particles did not interfere.
The scientists have christened their quantum computer «Jiuzhang», in reference to one of China's oldest mathematical scripts. Uninvolved researchers consider the result to be valid overall: Jian-Wei Pan has long been considered a leading expert in experiments with light particles. Among other things, he is the mastermind behind the quantum satellite Micius, with which Chinese scientists have set new records for the entanglement of light particles.
As for China's quantum computer experiment, detailed questions remain unanswered. On his blog, Scott Aaronson reports that Google, among others, has raised doubts about whether classic computers really perform as poorly on the task as thought.
What is already certain is that, unlike Google's quantum chip, the Chinese experimental setup isn't suitable for other computing tasks. To do this, one would have to measure the photons after each beam splitter and based on the result, change the rest of the course. If this eventually succeeds, light-based quantum computers are likely to become a serious competitor to the previously dominant quantum computing techniques.
Spektrum der Wissenschaft
We’re a partner of «Spektrum der Wissenschaft» (link in German) and strive to make scientific information better available to you. Follow Spektrum der Wissenschaft if you like the articles.
Experts from science and research report on the latest findings in their fields – competent, authentic and comprehensible.