Discovering Dynamical Error-Correcting Gate with Geometric Formalism and Machine Learning

In modern quantum information technology, how to protect and manipulate quantum information has been a key question in realizing reliable quantum algorithms and simulations. It is imperative to implement quantum operation at the bottom level with higher accuracy. There are plenty of work focused on quantum control and gate design in the past decades. However, designing a proper quantum gate with the practical device constraint while achieving certain goals is a challenging task. In this work, we try to combine geometric formalism for quantum control and physics-informed neural network. We show that the approach of machine learning is capable of discovering and optimizing dynamical error-correcting gates with practical constraints.