Simulating quantum measurements and quantum correlations

Abstract

This PhD thesis is focused on the quantum measurement simulability problem, that is, deciding whether a given measurement can be simulated when only a restricted subset of measurements is accessible. We provide an operational framework for this problem based on classical manipulations over the set of simulators. Particular cases of interest are further investigated, in which the simulators are taken to be projective measurements, measurements of a fixed number of outcomes, and arbitrary sets of fixed cardinality. In each of these situations we derive either necessary or sufficient conditions for simulability, and full characterisations in terms of semidefinite programming for some specific cases. Since joint measurability is a particular case of simulability, we also present a natural generalisation for it. Besides deciding whether a given measurement is simulable by some setof simulators, we also pose the question of what are the most robust measurements against simulability. We provide a strategy for approximating the set of quantum measurements based on relaxing the positivity constraint. This allows us to identify the most robust qubit measurement in terms of projective simulability, as well as the most incompatible sets of N measurements, for N = 1, . . . , 5, which notably are found to be always projective. By applying our simulability results in the context of Einstein-Podolsky-Rosen steering and Bell nonlocality we are able to construct improved and more general local models. Starting from models for a finite number of measurements we obtain the first general method for constructing local models for arbitrary families of quantum states. Similarly, our study on projective simulability yields a strategy for extending models for projective measurements to arbitraryones, culminating in the most efficient local model for two-qubit Werner states and general measurements.