Topologically noise robust network steering without inputs
Dhruv Baheti, Shubhayan Sarkar
TL;DR
The paper addresses observing quantum steering and nonlocality in networks without full inputs or strict independence assumptions by introducing swap-steering in triangle and general n-ring networks with a single trusted node. It develops linear witnesses to detect swap-steering, proves a separable-hidden-state bound of ${eta_{SOHS}=1/2}$, and shows a concrete quantum scheme achieving the algebraic maximum ${W=1}$ in the triangle, plus a scalable maximal violation for any n with the same strategy. A weak self-testing result certifies the states and measurements under plausible assumptions, and the framework is extended to topologically robust scenarios where untrusted parts may be arbitrarily connected, yet still exhibit a quantum advantage. Importantly, when the trusted party has tomography, every bipartite entangled state can generate swap-steerable correlations in the n-ring network, broadening the scope of detectable quantum correlations in networks. Collectively, these results provide a topologically robust, noise-tolerant route to network quantum advantage with reduced reliance on network structure and iid assumptions, with potential implications for secure quantum networking tasks.
Abstract
Quantum networks with independent sources allow observing quantum nonlocality or steering with just a single measurement per node of the network, or without any inputs. Inspired by the recently introduced notion of swap-steering, we consider here the triangle network scenario without inputs, where one of the parties is trusted to perform a well-calibrated measurement. In this scenario, we first propose a linear witness to detect triangle network swap-steering. Then, by using the correlations that achieve the maximum value of this inequality, and assuming that all the sources are the same, we can self-test the state generated by the sources and the measurements of the untrusted party. We then extend this framework to ring networks with an arbitrary number of nodes with one of them being trusted. Interestingly, this is the first example of a topologically robust, that is, one can observe steerability without assuming the network structure of the network, as well as noise-robust quantum advantage in a network. Additionally, by allowing the trusted party to perform tomography of their subsystems, we demonstrate that every bipartite entangled state will result in swap-steerable correlations in the ring network. For this purpose, we construct linear witnesses to detect ring network swap-steering corresponding to every bipartite entangled state.
