Year: 2024
Type: dissertation
Abstract: The objective behind developing quantum technologies is to perform computation and communication tasks that are infeasible using only classical resources. However, physical noise, due to interactions with the environment, and statistical noise, a result of the Born rule, are barriers to realizing the use cases of these technologies. In this thesis, we explore the effects of noise in a variety of quantum application settings and consider strategies to improve the performance of the protocols. A necessary step to executing useful tasks that require deep circuits is implementing error correction to handle faulty hardware. Motivated by a desire to inform near-term small circuit demonstrations of fault tolerance, we analyze the Bacon-Shor code and show that pseudo-thresholds are achievable on current hardware. Moreover, we demonstrate that a performance improvement can be realized by having error correction meta-parameters, such as the frequency of syndrome extraction, change as a function of the physical circuit structure and noise model. Next, we consider the effect of statistical noise on a key subroutine for variational quantum algorithms and design an adaptive algorithm for estimating quantum observables. This algorithm is shown to use information gained during the course of an experiment to outperform existing state-of-the-art methods, leading to an order of magnitude improvement in sample complexity, even for small problem instances. In the context of quantum networks, we study the effects of noise sources, such as photon loss, timing jitter, and dark counts, on dispersive-optics (DO) implementations of multi-party quantum key distribution (QKD). Using Gaussian quantum information extremality results, we provide achievable rate bounds for different noise models and a variety of network configurations (e.g. number of users in the network), helping inform the next generation of experiments and answering questions about how to scale these networks. Furthermore, we prove that the security analysis used in prior DO-QKD experiments is flawed in the finite-dispersion regime by providing a collective attack for the eavesdropper that has not been previously considered. We show that in realistic settings this results in a significant overestimation of the secure key rate.
Author: Shlosberg, Ariel S.
Institutions: Colorado School of Mines, University of Colorado System
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7047950222
