We investigate permutation-invariant continuous variable quantum states and their covariance matrices. We provide a complete characterization of the latter with respect to permutation-invariance, exchangeability and representing convex combinations of tensor power states. On the level of the respective density operators this leads to necessary criteria for all these properties which become necessary and sufficient for Gaussian states. For these we use the derived results to provide de Finetti-type theorems for various distance measures.
Entanglement is one of the most fascinating features arising from quantum-mechanics and of great importance for quantum information science. Of particular interest are so-called hybrid-entangled states which have the intriguing property that they contain entanglement between different degrees of freedom (DOFs). However, most of the current continuous variable systems only exploit one DOF and therefore do not involve such highly complex states. We break this barrier and demonstrate that one can exploit squeezed cylindrically polarized optical modes to generate continuous variable states exhibiting entanglement between the spatial and polarization DOF. We show an experimental realization of these novel kind of states by quantum squeezing an azimuthally polarized mode with the help of a specially tailored photonic crystal fiber.
We study the `local entanglement remaining after filtering operations corresponding to imperfect measurements performed by one or both parties, such that the parties can only determine whether or not the system is located in some region of space. The local entanglement in pure states of general bipartite multidimensional continuous-variable systems can be completely determined through simple expressions. We apply our approach to semiclassical WKB systems, multi-dimensional harmonic oscillators, and a hydrogen atom as three examples.
Kernel methods are ubiquitous in classical machine learning, and recently their formal similarity with quantum mechanics has been established. To grasp the potential advantage of quantum machine learning, it is necessary to understand the distinction between non-classical kernel functions and classical kernels. This paper builds on a recently proposed phase space nonclassicality witness [Bohmann, Agudelo, Phys. Rev. Lett. 124, 133601 (2020)] to derive a witness for the kernels quantumness in continuous-variable systems. We discuss the role of kernels nonclassicality in data distribution in the feature space and the effect of imperfect state preparation. Furthermore, we show that the non-classical kernels lead to the quantum advantage in parameter estimation. Our work highlights the role of the phase space correlation functions in understanding the distinction between classical machine learning from quantum machine learning.
In an abstract sense, quantum data hiding is the manifestation of the fact that two classes of quantum measurements can perform very differently in the task of binary quantum state discrimination. We investigate this phenomenon in the context of continuous variable quantum systems. First, we look at the celebrated case of data hiding against the set of local operations and classical communication. While previous studies have placed upper bounds on its maximum efficiency in terms of the local dimension and are thus not applicable to continuous variable systems, we tackle this latter case by establishing more general bounds that rely solely on the local mean photon number of the states employed. Along the way, we perform a quantitative analysis of the error introduced by the non-ideal Braunstein--Kimble quantum teleportation protocol, determining how much two-mode squeezing and local detection efficiency is needed in order to teleport an arbitrary local state of known mean energy with a prescribed accuracy. Finally, following a seminal proposal by Winter, we look at data hiding against the set of Gaussian operations and classical computation, providing the first example of a relatively simple scheme that works with a single mode only. The states employed can be generated from a two-mode squeezed vacuum by local photon counting; the larger the squeezing, the higher the efficiency of the scheme.
A sequence of random variables is exchangeable if its joint distribution is invariant under variable permutations. We introduce exchangeable variable models (EVMs) as a novel class of probabilistic models whose basic building blocks are partially exchangeable sequences, a generalization of exchangeable sequences. We prove that a family of tractable EVMs is optimal under zero-one loss for a large class of functions, including parity and threshold functions, and strictly subsumes existing tractable independence-based model families. Extensive experiments show that EVMs outperform state of the art classifiers such as SVMs and probabilistic models which are solely based on independence assumptions.