We use an error backpropagation (BP) neural network to determine whether an arbitrary two-qubit quantum state is steerable and optimize the steerability bounds of the generalized Werner state. Results show that, regardless of how we select the features for the quantum states, we can use the BP neural network to construct several models to obtain high-performance quantum steering classifiers compared with the support vector machine. Moreover, we predict the steerability bounds of the generalized Werner states using the classifiers that are newly constructed by the BP neural network; that is, the predicted steerability bounds are closer to the theoretical bounds. In particular, high-performance quantum steering classifiers with partial information about the quantum states that we need to measure in only three fixed measurement directions are obtained.

The study of quantum nonlocal correlations in networks (QNCNs), beyond the standard Bell theory with a single source, involves simulating large-scale quantum communications. The detectability of QNCNs depends on the input sources and selected measurements, and more essentially on the network structures. In this paper, we prove that same-size acyclic networks with different independence degrees can generate distinct sets of QNCNs in the device-independent scenario. We then distinguish two acyclic networks with different independence degrees by comparing the QNCNs generated from them. Furthermore, we show quantum demonstrations of the aforementioned notions in two-qubit-source networks. We investigate the quantitative influence of different network topologies on the detection of QNCNs in acyclic networks with Werner states as sources. Finally, we study how to maintain high independence degrees in networks formed by connecting two networks. This makes the correlations on the newly formed quantum network more likely to be nonlocal. Our research provides an alternative perspective for the well-known graph classification problem, and establishes an intrinsic connection between the graph theory and the foundation of quantum mechanics.

Sharing nonlocality refers to whether the postmeasurement state in a Bell test can be reused to demonstrate nonlocality among multiple observers conducting sequential measurements. It has become one of fascinating and challenging topics in the past decade. In this study, we shift the nonlocality sharing scenario to near the Schwarzschild black hole to explore how Hawking radiation affects sequential nonlocality sharing. In the ideal case without Hawking radiation, Alice and Bob share a maximally entangled pure state. Bob measures his half and then passes it to a second Bob, who repeats the measurement process, and so on. Previous studies have shown that Bob can perform an infinite number of measurements to achieve nonlocality sharing with Alice. If only Alice or Bob is affected by Hawking radiation, the number of shares becomes finite and depends on the Hawking temperature. Unfortunately, when both are exposed to Hawking radiation, sharing nonlocality becomes nearly impossible. However, we can overcome this limitation by introducing auxiliary entangled sources. By adding several observers sharing an entangled state with Alice, we form a star network. Our results indicate that with these auxiliary sources, sequential network nonlocality sharing is achievable within this star network.

We transfer the scenario of quantum network nonlocal correlation to the vicinity of a black hole’s event horizon, aiming to evaluate the impact of Hawking effect on network nonlocality within this specific context. In an entanglement swapping network with two bipartite resources and three parties, we examine two scenarios with distinct Hawking radiation locations: one where it affects the endpoint parties’ system and another where it impacts all parties’ system. In both cases, network nonlocality decays due to Hawking radiation and disappears entirely when the Hawking temperature reaches a critical threshold. However, we observe a striking phenomenon: when resource states are maximally entangled and the affected system belongs to the endpoint participants, network nonlocality diminishes but does not vanish throughout the radiation process. Subsequently, we extend our investigation to star and chain networks to evaluate how resource quantity and distribution influence network nonlocality under Hawking radiation. Our results indicate that star networks exhibit greater resilience against Hawking radiation in maintaining nonlocality.