A breakthrough study led by the University of Oxford has utilized machine learning to bridge the reality gap in quantum devices, a significant challenge in the field of quantum computing. The findings, published in Physical Review X, offer a promising solution to the inherent variability that hampers the scaling and combination of quantum devices, known as qubits.
Quantum computing has the potential to revolutionize various domains, including climate modeling, financial forecasting, drug discovery, and artificial intelligence. However, the variability exhibited by seemingly identical quantum devices has hindered their effective utilization. This variability is believed to be caused by nanoscale imperfections in the materials used to construct these devices. As there is currently no direct method to measure these imperfections, the internal disorder cannot be accurately captured in simulations, resulting in a discrepancy between predicted and observed outcomes.
To address this challenge, the research team adopted a physics-informed machine learning approach to indirectly infer the characteristics of internal disorder. They focused on how the flow of electrons through the device is affected by the internal disorder. By analyzing the output current at different voltage settings across a quantum device, the researchers inputted the data into a simulation that calculated the difference between the measured current and the theoretical current in the absence of internal disorder. The simulation was then constrained to identify an arrangement of internal disorder that could explain the measurements at all voltage settings. This approach combined mathematical and statistical techniques with deep learning.
The new model not only successfully identified suitable internal disorder profiles to describe the measured current values but also accurately predicted the voltage settings required for specific device operating regimes.
Significantly, this model provides a novel means to quantify the variability between quantum devices. It has the potential to enhance predictions of device performance and guide the engineering of optimal materials for quantum devices. Moreover, it could assist in developing compensation strategies to mitigate the adverse effects of material imperfections in quantum devices.
David Craig, co-author and a PhD student at the Department of Materials, University of Oxford, likened their approach to indirectly inferring the presence of black holes by observing their impact on surrounding matter. Though the real quantum devices possess greater complexity than what the model captures, the study demonstrates the value of utilizing physics-aware machine learning to narrow the reality gap.
This breakthrough has opened up new possibilities for harnessing the power of quantum computing and could pave the way for advancements in various fields. The ability to overcome the challenges posed by device variability holds promise for scaling and combining quantum devices and unlocking their full potential. With further research and development, quantum computing may soon revolutionize industries and accelerate scientific discoveries.
