Quantum bounds and device-independent security with rank-one qubit measurements
Device-independent (DI) quantum protocols exploit Bell inequality violations to ensure security or certify quantum properties without making assumptions about the internal workings of the devices. In this work, we study the role of rank-one qubit positive operator-valued measures (POVMs) in DI scenarios. This class includes all qubit extremal POVMs, i.e., those measurements that cannot be realized by randomly choosing among others, as well as part of non-extremal POVMs, which have recently been shown to be useful for security applications in sequential quantum protocols. We demonstrate that any rank-one POVM can generate correlations in bipartite scenarios that saturate a Tsirelson inequality, i.e., a quantum bound on linear combinations of outcome statistics, when the two parties share an arbitrary entangled two-qubit state and some other self-tested measurements are performed. For extremal POVMs, such saturation allows for an explicit calculation of the guessing probability and the worst-case conditional von Neumann entropy. From the Tsirelson inequality, we establish a randomness certification method that facilitates numerical simulations and noise analysis. To test its feasibility, we performed a proof-of-concept experiment employing a three-outcome POVM on tilted entangled states under experimental non-idealities. We further explore the case of non-extremal POVMs, providing insights into their role in DI protocols.