Miniaturizing neural networks for charge state autotuning in quantum\n dots

Abstract

A key challenge in scaling quantum computers is the calibration and control\nof multiple qubits. In solid-state quantum dots, the gate voltages required to\nstabilize quantized charges are unique for each individual qubit, resulting in\na high-dimensional control parameter space that must be tuned automatically.\nMachine learning techniques are capable of processing high-dimensional data -\nprovided that an appropriate training set is available - and have been\nsuccessfully used for autotuning in the past. In this paper, we develop\nextremely small feed-forward neural networks that can be used to detect\ncharge-state transitions in quantum dot stability diagrams. We demonstrate that\nthese neural networks can be trained on synthetic data produced by computer\nsimulations, and robustly transferred to the task of tuning an experimental\ndevice into a desired charge state. The neural networks required for this task\nare sufficiently small as to enable an implementation in existing memristor\ncrossbar arrays in the near future. This opens up the possibility of\nminiaturizing powerful control elements on low-power hardware, a significant\nstep towards on-chip autotuning in future quantum dot computers.\n