Breaking adiabatic quantum control with deep learning

In the noisy intermediate-scale quantum era, optimal digitized pulses are requisite for efficient control. This goal is translated into dynamic programming, in which a deep reinforcement learning (DRL) agent gifted. As reference, shortcuts to adiabaticity (STA) provide analytical approaches adiabatic speedup by pulse Here, we select single-component control of qubits, resembling ubiquitous two-level Landau-Zener problem gate operation. We aim at obtaining fast and robust digital combining STA DRL algorithm. particular, find that leads with operation time bounded speed limits dictated STA. addition, demonstrate robustness against systematic errors can be achieved without any input from Our results introduce general framework control, leading promising enhancement information processing.