Abstract: The ability to quickly and accurately characterize quantum states and dynamics is crucial for the development of quantum technologies. However, the problem of learning a general quantum state or process has exponential complexity in the size of the quantum system. In this talk, I will present some recent progress we have made for both quantum state and process tomography. For state tomography, I will show how generative adversarial neural networks can outperform standard methods in terms of both the amount of time and data needed [1,2]. For process tomography, I will show how optimization using constrained gradient descent can work both for instances with little data and for larger systems, regimes which previously required two different methods [3].
[1] S. Ahmed et al., Phys. Rev. Lett. 127, 140502 (2021)
[2] S. Ahmed et al., Phys. Rev. Res. 3, 033278 (2021)
