Spectra of noisy parameterized quantum circuits: Single-Ring universality

Random unitaries are an important resource for quantum information processing. While their universal properties have been thoroughly analyzed, it is not known what happens to these properties when the unitaries are sampled on the present-day noisy intermediate-scale quantum (NISQ) computers. We implement parameterized circuits, which have been proposed as a means to generate random unitaries, on an IBM Quantum processor and model these implementations as quantum maps. To retrieve the maps, a machine-learning assisted tomography is used. We find the spectrum of a map to be either an annulus or a disk depending on the circuit depth and detect an annulus-disk transition. By their spectral properties, the retrieved maps appear to be very similar to a recently introduced ensemble of random maps, for which spectral densities can be analytically evaluated. Our results establish, via Dissipative Quantum Chaos theory, a connection between intrinsic properties of present-day noisy intermediate-scale quantum (NISQ) computing platforms and non-Hermitian random matrix theory.