Tuning Arrays with Rays: Physics-Informed Tuning of Quantum Dot Charge States

Abstract

Many methods to automatically tune silicon spin qubits are limited by reliability and data efficiency, which makes them less likely to be scalable. The authors demonstrate a reliable, efficient, physics-informed tuning algorithm (PIT) for navigating to a target charge configuration⏤a prerequisite to forming qubits. This tuning method combines machine learning and physical intuition with an algorithm that leverages one-dimensional scans (rays) and conventional peak-finding to navigate from a coarse, unknown device state to a desired charge occupation efficiently and effectively. PIT enables the transformation of an uncalibrated circuit to a functioning quantum processor.