Experimental Online Quantum Dots Charge Autotuning Using Neural Network

• DOI: 10.48550/arxiv.2409.20320
• Publication Date: 2024-09-30
• Authors: Victor Yon; Bastien Galaup; Claude Rohrbacher; Joffrey Rivard; A. Morel; Dominic Leclerc; C. Godfrin; Ruoyu Li; Stefan Kubicek; Kristiaan De Greve; Eva Dupont-Ferrier; Yann Beilliard; Roger G. Melko; Dominique Drouin
• Keywords: Quantum Physics; I.5.1; 81V65 (Primary), 68T37 (Secondary); Condensed Matter - Mesoscale and Nanoscale Physics; I.2.8; Mesoscale and Nanoscale Physics (cond-mat.mes-hall); FOS: Physical sciences; Quantum Physics (quant-ph); I.2.8; I.5.1

Spin-based semiconductor qubits hold promise for scalable quantum computing, yet they require reliable autonomous calibration procedures. This study presents an experimental demonstration of online single-dot charge autotuning using a convolutional neural network integrated into closed-loop system. The algorithm explores the gates' voltage space to localize transition lines, thereby isolating one-electron regime without human intervention. In 20 runs on device cooled 25mK, method achieved success rate 95% in locating target electron regime, highlighting robustness this against noise and distribution shifts from offline training set. Each tuning run lasted average 2 hours 9 minutes, primarily due limited speed current measurement. work validates feasibility machine learning-driven real-time dot devices, advancing development toward control large qubit arrays.