Data Generation for Triple-Quantum-Dot Machine-Learning Autotuning

Tuning quantum dot devices is a time-consuming task which becomes increasingly difficult for larger devices. However, machine-learning techniques have the potential to automate the device-tuning process. We use a Thomas-Fermi approximation to simulate data for triple-dot devices, which is then used to train a machine-learning model. This model can determine the region of parameter-space that a device is in (which dots are empty / filled / overfilled), using only a small number of local measurements. This information can then be used to determine how gate voltages need to be adjusted to reach the desired charge state. The classification of data using machine-learning tools is a pivotal step towards fully automating the process of tuning quantum dot devices.