Testing the limits of nonlocality

The nascent field of quantum information offers the promise of dramatic speed increases for certain types of computation and logical protocols which are classically impossible. The phenomenon of nonlocality is fundamental to this study, and polarization-entangled photons o er a clean, bright, and stable system for its experimental investigation. This thesis investigates, both experimentally and theoretically, the creation, manipulation, and measurement of polarization-entangled photon pairs. Our entangled photon pairs are created through spontaneous parametric down-conversion within a pair of adjacent, orthogonally oriented nonlinear crystals. The quality and brightness of this source is then dramatically improved through the use of birefringent crystals which compensate for an angle-dependent phase shift. The use of these compensation crystals has allowed measured count rates of two million pairs per second with 97.7% fidelity (with a maximally entangled state) or alternately, ten thousand pairs per second with 99.5% fidelity. By manipulating these entangled states, it is possible to study both how they change and how they can be used. We discuss the theory of state manipulation and the experimental implementation of extremely precise single-qubit operations. To study the operations themselves, we use quantum process tomography to characterize them and have successfully implemented the first experimental realization of ancilla-assisted process tomography. To study decoherence, we implemented the first experimental decoherence-free subspace. By using techniques developed during these investigations, we are able to transform our source of entangled photons into a source of any two-photon polarization state, mixed or pure. Developing new techniques for state creation and manipulation is possible because of simultaneous development in state measurement. In addition to detailing both a theoretical analysis and experimental instructions for state tomography, we experimentally and theoretically compared state tomography with tests of local realism (Bell inequalities) and entanglement witnesses (an entanglement detection technique). In the process, we have measured the largest violations of local realism to date, both statistical (over 2400-sigma) and absolute (S = 2:826 ± 0:005 -- within 0:2% of a maximal violation).U of I Onlydissertation/thesi