B and F are a quantum system’s input and output states, respectively. E is an auxiliary system that allows the sequence of input state B to be passed to the quantum reservoir S. S can then read to imitate F without disrupting it. Credit: (c)2021 Tran et al.

Shortly, technologies that exploit novel quantum mechanical behavior will be commonplace. Devices that use quantum information to input and output data may be included. These devices will require careful verification because of inherent uncertainties. Warranty is more complicated if the device’s output is dependent on previous inputs. Researchers used machine learning to improve the efficiency of time-dependent quantum device verification dramatically. They did this by including a memory effect in these systems.

Although quantum computers are making headlines in scientific media, most experts still consider them to be in their infancy. However, a quantum internet may be closer to the future. This would provide significant security benefits over the current internet. However, even this will depend on technologies that have yet to see the outside world. Although many of the fundamentals of devices that will create our quantum internet have been figured out, many engineering problems remain in making these products. However, a lot of research is underway to develop tools that can be used in the design and construction of quantum devices.

Quoc Hoantran, a postdoctoral researcher, and Kohei Nakajima, an Associate Professor at the Graduate School of Information Science and Technology in Tokyo, have created such a tool. It could make verifying quantum device behavior more efficient and accurate than it currently is. The algorithm they developed can reconstruct quantum devices’ time-dependent workings by simply learning the relationship between the quantum outputs and inputs. Although this approach is common when studying a classical system of physical systems, it is impossible to store quantum information, which is why it is often impossible.

Tran said that quantum process tomography describes quantum systems based on their inputs and outputs. However, many quantum system researchers have discovered that they exhibit a memory effect in which present states are affected by the previous. It is impossible to describe the system’s time dependence by simply inspecting its input and output states. It would be computationally inefficient to model the system multiple times after each change in time. We wanted to embrace the memory effect and use it rather than using brute force to overcome it.

Tran and Nakajima used machine learning and a technique known as quantum reservoir computing to create their algorithm. It learns patterns in inputs and outputs that change over time in quantum systems and predicts how they will change in future situations. The algorithm is more straightforward than an empirical method and produces faster results.

“At the moment, our algorithm can emulate certain types of quantum systems, but hypothetical devices might vary in their processing abilities and memory effects. Tran stated that the next step in the research would be to expand the capabilities of our algorithms. This will make something more useful and, therefore, more efficient. “I’m excited about what quantum machine-learning methods can do and the potential devices they might lead to.”