Researchers have developed metaQctrl, a two-layer meta-reinforcement learning algorithm that significantly outperforms conventional methods in achieving high-fidelity quantum gates with fewer control pulses under uncertain conditions, potentially advancing practical quantum computing by maintaining 99.99% fidelity where traditional approaches fail.
It’s easy to control the trajectory of a basketball: all we have to do is apply mechanical force coupled with human skill. But controlling the movement of quantum systems such as atoms and electrons is much more challenging, as these minuscule scraps of matter often fall prey to perturbations that knock them off their path in unpredictable ways. Movement within the system degrades — a process called damping — and noise from environmental effects such as temperature also disturbs its trajectory.
Researchers trained a machine learning tool to capture the physics of electrons moving on a lattice using far fewer equations than would typically be required, all without sacrificing accuracy.
German Federal Ministry of the Economy has funded the €19m PlanQK project (Platform and Ecosystem for Quantum-Assisted Artificial Intelligence) in Quantum AI.