资源环境科技发展态势分析平台

Although they rely on the amazing properties of quantum particles, quantum computers do not systematically outperform today's machines. The race for quantum supremacy aims to find applications in which even less powerful quantum computers can far exceed the performance of the best "conventional" supercomputers. In 2019, Google claimed to be the first to have achieved quantum supremacy, which its competitor IBM immediately challenged, before a team from the University of Science and Technology of China followed suit. In an article published on 8 February 2021 in Nature Communications, researchers from the CNRS, the University of Edinburgh (UK), and the company QC Ware have joined this exclusive circle.

In a field that is prone to a growing number of announcements and relentless competition among major powers, definitions and achievements must always be clarified: "Concerning our research, I prefer the term algorithmic quantum advantage to quantum supremacy," says Eleni Diamanti, co-author of the study and senior researcher at the LIP6.1 "The machine designed by Google can only perform a particular task so far, although it is meant to evolve towards a universal quantum computer, namely one that can perform all sorts of calculations. Our instrument is closer to that of the Chinese team, in other words a processor that is adapted to very specific situations."

Towards quantum cloud computing

The scientists focused on quantum computers that will be accessible via the cloud, which should become their most common form of use in coming years. "Just as we don't have a supercomputer at home, these machines will mainly be available online to start with," explains Iordanis Kerenidis, co-author of the study, senior researcher at the Research Institute on the Foundations of Computer Science (IRIF),2 and head of the Quantum Algorithms International department at QC Ware. "So what we need are genuine quantum networks that are both effective and secure, in order to establish a reliable link between providers and their clients."

Such a service will in particular have applications in logistics and learning models for artificial intelligence, in which many problems still elude conventional computers. Once again, it is a matter of proving that the quantum system is much more effective. Researchers reached quantum advantage with their algorithm, which actually proves the quantum advantage... of other algorithms.

"In a matter of seconds, we can verify whether a system can outperform a traditional computer," enthuses Kerenidis. "Our algorithmic solution can determine whether it has the ability to come up with a solution to the problem, without having to solve it. In the event the provider does not want to give clients the entire solution, it would take an ordinary computer as long as the age of the Universe to do so."

A question of imprints

"The core of our research was to innovate in order to find the right algorithm," Diamanti concludes. "We then put it into use, proving its quantum advantage through a combination of theory and experience. To do so, we needed to run our algorithm on a system that enabled us to demonstrate quantum advantage despite experimental errors. We resorted to quantum states known as ‘imprints’, i.e. the information given by the provider to its clients."

Yet what kind of machine does the algorithm operate on to obtain these results? Researchers speak of a quantum processor, which is more specialised than a quantum computer: the system functions with photons, which carry qubits generated by a laser and manipulated by an optic circuit. The light is then detected, and the results of the detection are processed in order to solve the problem. The equipment is less sophisticated than the solutions proposed by Google and China, and therefore has the advantage of being reproducible in most cutting-edge optics laboratories.

In search of new solutions

"Our interdisciplinary research involves collaboration between physicists and computer scientists from the CNRS," Kerenidis stresses. "They are part of the Paris Centre for Quantum Computing  (PCQC), which Eleni Diamanti and I jointly head." At the company level, QC Ware is looking for quantum applications that use the least complicated and resource-intensive machines as possible. This is a crucial condition, for the scientists could have designed their system using thousands of lasers and sensors, but strove to reduce their use to a minimum.

The team is now pursuing its search towards new applications that can demonstrate quantum advantage. Diamanti has specialised in cryptography adapted to the quantum world, while Kerenidis is focusing on machine learning. They will also study this approach in connection with data management and big data, where quantum technology promises huge improvements.