A team of researchers from the University of Oxford has made significant progress in addressing a long-standing challenge in quantum devices through the use of machine learning. Quantum devices, which have the potential to revolutionize various fields such as climate modeling and drug discovery, often suffer from inherent variability, where seemingly identical units exhibit different behaviors. This variability is thought to be caused by nanoscale imperfections in the materials of the devices.
In an effort to bridge the gap between predicted and observed behavior in quantum devices, the research team employed a physics-informed machine learning approach. By analyzing the flow of electrons through the devices and inferring the internal disorder patterns indirectly, they were able to close the ‘reality gap’. This approach, likened to playing “crazy golf” where the ball’s movements can be predicted with practice and data collection, allowed the researchers to make more accurate predictions about the devices’ performance.
The researchers measured the output current of individual quantum dot devices at different voltage settings and used this data to constrain a simulation. The simulation calculated the difference between the measured current and the theoretical current without internal disorder, allowing the researchers to find suitable internal disorder profiles that could explain the measurements. This combination of mathematical, statistical, and deep learning approaches proved effective in predicting voltage settings for specific device operating regimes.
Furthermore, the new model developed by the research team provides a means to quantify the variability between quantum devices. This advancement could lead to more accurate predictions of device performance and aid in the engineering of optimal materials for quantum devices. The model also offers insights into compensation approaches to mitigate the effects of material imperfections.
Overall, this study represents a significant step forward in harnessing the power of machine learning to overcome the obstacles posed by variability in quantum devices. With further research and development, this approach could contribute to the widespread adoption and utilization of quantum computing technology in various industries.
The source of the article is from the blog elperiodicodearanjuez.es