Quantum Coherence Lab

Zumbühl Group


New ArXiv submission: Effciently measuring a quantum device using machine learning

Tuning quantum devices is becoming time-consuming as systems are scaled up, e.g. to numerous gates or contacts, and will soon become intractable without automation. Here, we present measurements on a quantum dot done by a machine learning algorithm. This selects the most informative measurements to perform next using information theory and a probabilistic deep-generative model capable of generating multiple full-resolution reconstructions from scattered partial measurements. We demonstrate that the algorithm outperforms standard grid scan techniques, reducing the measurement time by a factor of ~4,  thus laying the foundation for automated control of large quantum circuits. Lennon et al., arxiv.org/abs/1810.10042