Francesco Anna Mele, Antonio Anna Mele, Lennart Bittel, Jens Eisert, Vittorio Giovannetti, Ludovico Lami, Lorenzo Leone, Salvatore F.E. Oliviero (May 03 2024).
Abstract: Quantum state tomography with rigorous guarantees with respect to the trace distance, the most operationally meaningful metric for distinguishing quantum states, has been studied extensively for finite-dimensional systems; however, it remains almost unexplored for continuous variable systems. This work fills this gap. We prove that learning energy-constrained $n$-mode states without any additional prior assumption is extremely inefficient: The minimum number of copies needed for achieving an $\varepsilon$-approximation in trace distance scales as $\sim \varepsilon^{-2n}$, in stark contrast to the $n$-qudit case, where the $\varepsilon$-scaling is $\sim \varepsilon^{-2}$. Specifically, we find the optimal sample complexity of tomography of energy-constrained pure states, thereby establishing the ultimate achievable performance of tomography of continuous variable systems. Given such an extreme inefficiency, we then investigate whether more structured, yet still physically interesting, classes of quantum states can be efficiently tomographed. We rigorously prove that this is indeed the case for Gaussian states, a result previously assumed but never proved in the literature. To accomplish this, we establish bounds on the trace distance between two Gaussian states in terms of the norm distance of their first and second moments, which constitute technical tools of independent interest. This allows us to answer a fundamental question for the field of Gaussian quantum information: by estimating the first and second moments of an unknown Gaussian state with precision $\varepsilon$, what is the resulting trace distance error on the state? Lastly, we show how to efficiently learn $t$-doped Gaussian states, i.e., states prepared by Gaussian unitaries and at most $t$ local non-Gaussian evolutions, unveiling more of the structure of these slightly-perturbed Gaussian systems.