Speaker
Description
Studying the boundary between quantum and classical physics in macroscopic systems is a central challenge in modern physics. In this context, levitated systems are becoming increasingly important, as they provide excellent isolation and avoid clamping losses, enabling improved quantum control of mechanical motion compared to traditional micro- and nanomechanical resonators. Beyond optical tweezers, magneto-levitated systems based on static superconducting traps that exploit the Meissner effect are particularly promising, as they operate at cryogenic temperatures and exhibit exceptional immunity to dissipative effects. Achieving quantum superposition of massive objects in magnetically levitated systems requires cooling their motion to the quantum ground state. In this talk, I will present a recent proof-of-concept experiment in which we cooled the angular motion of a levitated micromagnet in a superconducting Meissner trap using magnetic feedback. We analyzed the experimental data with a theoretical model tailored to our feedback scheme. Starting from an operating temperature of a few kelvin, we cooled the system down to a few millikelvin. We also discuss how lower temperatures and ground state cooling could be reached through improvements in the performance of the experimental apparatus.
Finally, I will discuss preliminary results from a theoretical study exploring the use of a machine-learning-based strategy (reinforcement learning) to enhance the feedback cooling process. This approach introduces an adaptive agent that controls the feedback in real time, with the ultimate goal of reaching the quantum regime of a massive levitated system.