China quantum research team uses AI to crack processing time in supercomputer race
Artificial intelligence is set to transform quantum research with a huge boost in processing speed that could ultimately pave the way for a supercomputer a billion " or even a trillion " times faster than existing technology.
According to a new study by scientists in China " which has established some of the world's largest quantum research facilities " machine learning technology can slash the time-consuming task of calculating the nature of the quirky ties between particles that exist in the subatomic quantum realm.
The power of quantum technology arises from particles that are linked by the mysterious forces of quantum mechanics.
Quantum entanglement, for example, refers to the correlation between particles that allows changes in one to instantly affect another, regardless of the distance between them. The behaviour of these correlated particles forms the basis of quantum information processing.
The power of quantum technology arises from particles that are linked by the mysterious forces of quantum mechanics. Photo: Shutterstock alt=The power of quantum technology arises from particles that are linked by the mysterious forces of quantum mechanics. Photo: Shutterstock
Now, a Chinese research team has established that machine learning technology can assess the nature of the ties between particles with unprecedented speed and accuracy, according to a paper published this month in the scientific journal Physical Review Letters.
This is important, according to the researchers, because different ties between particles make them suitable for different tasks. As well as quantum entanglement, particles may be related through EPR steering " the paradoxical phenomenon best known for Schrodinger's Cat " or Bell nonlocality " which states that in some situations the property of a particle can be altered by the act of measuring it.
Most quantum devices " from key distribution networks for ultra-safe communication, superfast quantum computers, to quantum radar systems for stealth aircraft detection " remain tantalisingly beyond practical application because of the huge amount of traditional processing time required now to establish the nature of the ties between particles.
"It works like the separation of minerals," said Professor Li Chuanfeng, a lead scientist of the study with the University of Science and Technology of China (USTC) in Hefei, Anhui province.
"The AI tells us whether an ore contains gold, iron or copper, so we can make use of them for different purposes. This was not possible before," Li said.
Until now, quantum researchers have had to measure a complete set of physical properties to determine the type of link between particles. It is a difficult, time consuming job and, when the number of particles increases, the workload grows exponentially.
"At the end of the day we might be able to tell this is not a piece of gold, and that would be it. There is no way to know more," Li said.
USTC researchers have been conducting quantum experiments for decades and amassed a large amount of data. Li's team built up a data set containing more than 400 pairs of particles carefully picked with different types of links, such as quantum entanglement, EPR steering and Bell nonlocality.
The scientists then used the data set to train a computer with a deep learning algorithm to measure the physical properties of the particles. The AI can now achieve an accuracy higher than 90 per cent, and the time taken has been slashed to the point where calculations that previously took an hour can be worked out in a fraction of a second.
The machine also does not need full information on each particle to make the assessment. According to the paper, researchers only need to feed details of two physical properties into the machine, which can then fill in the gaps and make a correct estimate on the outcome.
"This method will increase the supply of resources for quantum information processing significantly," Li said.
Ren Changliang, associate professor with the Chongqing Institute of Green and Intelligent Technology and co-author of the paper, said the breakthrough did not mean AI had a better grasp on the world of quantum physics than the human brain.
"We label the data, we teach it, we correct its mistakes. The AI follows human guidance. It is not outsmarting us," he said.
Another obvious advantage of AI over traditional methods is that it can deal with multidimensional issues much more efficiently " and there are many dimensions in quantum physics.
"There seems to be a good match between the two worlds," Ren said.
According to the scientists, this is not the end of their research. The team is now planning to continue training the machine with a larger data set, and is also developing new artificial intelligence technology specifically for a quantum computer, which is predicted to be up to 1 trillion times faster than the most powerful supercomputer today.
Some researchers say that a combination of AI and quantum computing might eventually lead to machines with intelligence equal or greater to that of humans.
This article originally appeared in the South China Morning Post (SCMP), the most authoritative voice reporting on China and Asia for more than a century. For more SCMP stories, please explore the SCMP app or visit the SCMP's Facebook and Twitter pages. Copyright © 2019 South China Morning Post Publishers Ltd. All rights reserved.
Copyright (c) 2019. South China Morning Post Publishers Ltd. All rights reserved.
