Design of optimization tools for quantum information theory
In this thesis, we present optimization tools for different problems in quantum information theory. First, we introduce an algorithm for quantum estate estimation. The algorithm consists of orthogonal projections on intersecting hyperplanes, which are determined by the probability distributions and the measurement operators. We show its performance, in both runtime and fidelity, considering realistic errors. Second, we present a technique for certifying quantum non-locality. Given a set of bipartite measurement frequencies, this technique finds a Bell inequality that maximizes the gap between the local hidden variable and the quantum value of a Bell inequality. Lastly, to study the quantum marginal problem, we introduce an operator and develop an algorithm, which takes as inputs a set of quantum marginals and eigenvalues, and outputs a density matrix, if exists, compatible with the prescribed data.