Super-Resolution Reconstruction of Continuous-Variable Quantum Systems

Full characterization of continuous-variable quantum systems is traditionally resource-intensive, requiring extensive measurements and post-processing. To address this challenge, we present a super-resolution method that learns smooth Wigner functions within the entire phase space from sparse and noisy data. This method is inspired by neural network-based super-resolution imaging. We test our method on various simulated scenarios, including bosonic code states with almost 100 photons and complex quantum dynamics, as well as real experimental data on superconducting experiments. The interpolated Wigner function can be used to predict various quantum properties. The results demonstrate that this neural network approach provides a highly efficient and accurate method for continuous-variable quantum system characterization.