Quantum computation: algorithms and implementation in quantum dot devices.

摘要

In this thesis, we explore several aspects of both the software and hardware of quantum computation. First, we examine the computational power of multi-particle quantum random walks in terms of distinguishing mathematical graphs. We study both interacting and non-interacting multi-particle walks on strongly regular graphs, proving some limitations on distinguishing powers and presenting extensive numerical evidence indicative of interactions providing more distinguishing power. We then study the recently proposed adiabatic quantum algorithm for Google PageRank, and show that it exhibits power-law scaling for realistic WWW-like graphs.;Turning to hardware, we next analyze the thermal physics of two nearby 2D electron gas (2DEG), and show that an analogue of the Coulomb drag effect exists for heat transfer. In some distance and temperature, this heat transfer is more significant than phonon dissipation channels. After that, we study the dephasing of two-electron states in a single silicon quantum dot. Specifically, we consider dephasing due to the electron-phonon coupling and charge noise, separately treating orbital and valley excitations. In an ideal system, dephasing due to charge noise is strongly suppressed due to a vanishing dipole moment. However, introduction of disorder or anharmonicity leads to large effective dipole moments, and hence possibly strong dephasing.;Building on this work, we next consider more realistic systems, including structural disorder systems. We present experiment and theory, which demonstrate energy levels that vary with quantum dot translation, implying a structurally disordered system. Finally, we turn to the issues of valley mixing and valley-orbit hybridization, which occurs due to atomic-scale disorder at quantum well interfaces. We develop a new theoretical approach to study these effects, which we name the disorder-expansion technique. We demonstrate that this method successfully reproduces atomistic tight-binding techniques, while using a fraction of the computational resources and providing considerably more physical insight. Using this technique, we demonstrate that large dipole moments can exist between valley states in disordered systems, and calculate corrections to intervalley tunnel rates..