Technologies that take advantage of new quantum mechanical behavior may become commonplace in the near future. These may include devices that use quantum information as input and output data and require careful verification due to the inherent uncertainty. Verification becomes more difficult if the device is time dependent when the output is dependent on past inputs. For the first time, researchers using machine learning have dramatically improved the efficiency of validation of time-dependent quantum devices by incorporating certain memory effects that exist in these systems.
Quantum computers have become headlines in scientific coverage, but these computers are considered by most experts to be still in their infancy. However, the quantum internet may be a little closer to the present. This offers significant security advantages, especially over the current Internet. But it still relies on technology that still can’t see the light of day outside the lab. Many of the basics of the devices that can create our quantum Internet may have been elucidated, but there are many engineering challenges in achieving these as products. However, much research is underway to create tools for the design of quantum devices.
Postdoctoral fellows at the Graduate School of Information Sciences, the University of Tokyo, Quoc Juan Trang and Associate Professor Kohei Nakajima have pioneered just such a tool. It is now. Their contribution is an algorithm that can reconstruct the behavior of time-dependent quantum devices simply by learning the relationship between quantum inputs and outputs. This approach is practically common when exploring classical physical systems, but it is usually not possible because quantum information is generally difficult to store.
“The technique of describing quantum systems based on inputs and outputs is called quantum process tomography,” says Tran. “But many researchers now report that their quantum system exhibits some kind of memory effect when the current state is affected by the previous state. This is the input state and output. A simple check of the state means that the time dependence of the system cannot be explained. It is possible to model the system iteratively with each time change, but this is very computationally inefficient. Our aim was to accept this memory effect and use it to our advantage, rather than using Brute Force to overcome it. “
Tran and Nakajima turned to a technique called machine learning and quantum reservoir computing to build new algorithms. It learns input and output patterns that change over time in a quantum system and effectively infers how these patterns change, even in situations where the algorithm has not yet witnessed it. Team algorithms are simpler and can produce results faster because they only need to know the inputs and outputs, not the internal behavior of the quantum system as in the more empirical method.
“Currently, our algorithms can emulate certain types of quantum systems, but virtual devices can be very different in processing power and memory effect. Therefore, the next step in research is the functioning of the algorithm. To make something more versatile and more useful in essence, “says Tran. “I’m excited about what quantum machine learning methods can do and the virtual devices they can bring.”
Reference: “Learning Time Quantum Tomography” December 22, 2021 Physical review letter..
DOI: 10.1103 / PhysRevLett.127.260401
New machine learning algorithms enable efficient and accurate verification of quantum devices
https://scitechdaily.com/a-novel-machine-learning-algorithm-allows-for-efficient-and-accurate-verification-of-quantum-devices/ New machine learning algorithms enable efficient and accurate verification of quantum devices