We propose a gradient-based general computational framework for optimizing model-dependent parameters in quantum batteries (QB). We apply this method to two different charging scenarios in the micromaser QB and we discover a charging protocol for stabilizing the battery in upper-laying Hilbert space chambers in a controlled and automatic way. This protocol is found to be stable and robust, and it leads to an improved charging efficiency in micromaser QBs. Moreover, our optimization framework is highly versatile and efficient, holding great promise for the advancement of QB technologies at all scales.
@article{arxiv.2301.09408,
title = {Artificial intelligence discovery of a charging protocol in a micromaser quantum battery},
author = {Carla Rodríguez and Dario Rosa and Jan Olle},
journal= {arXiv preprint arXiv:2301.09408},
year = {2023}
}