School of Physics - Theses

Permanent URI for this collection

Search Results

Now showing 1 - 6 of 6
  • Item
    Thumbnail Image
    Simulating Noisy Quantum Algorithms and Low Depth Quantum State Preparation using Matrix Product States
    Nakhl, Azar Christian ( 2021)
    Since the proposal of Quantum Computation in the 1980s, many Quantum Algorithms have been proposed to solve problems in a wide variety of fields. However, due to the limitations of existing quantum devices, analysing the performance of these algorithms in a controlled manner must be performed classically. The leading technique to simulate quantum computers classically is based on the Matrix Product State (MPS) representation of quantum systems. We used this simulation method to benchmark the noise tolerance of a number of quantum algorithms including Grover’s Algorithm, finding that the algorithm’s ability to discern the marked state is exponentially suppressed under noise. We verified the existence of Noise-Induced Barren Plateaus (NIBPs) in the Quantum Approximate Optimisation Algorithm (QAOA) and found that the recursive QAOA (RQAOA) variation is resilient to NIBPs, a novel result. Also integral to the performance of quantum algorithms is the ability to efficiently prepare their initial states. We developed novel techniques to prepare low-depth circuits for slightly entangled quantum states using MPS. We found that we can reproduce Gaussian and W States with circuits of O(log(n)) depth, improving on current best known results which are of O(n).
  • Item
    Thumbnail Image
    Moments-Based Corrections to Variational Quantum Computation
    Jones, Michael Alexander ( 2020)
    Quantum Computing offers the potential to efficiently solve problems for which there are no known, efficient classical solutions such as factoring of semi-prime numbers and simulation of quantum- mechanical systems. This work considers a novel moments-based adaptation of the Variational Quantum Eigensolver (VQE), one of the leading candidates for demonstrating quantum supremacy. The method for improving the estimated ground state energy of a quantum system, obtained using the Variational Quantum Eigensolver, is presented and tested for Heisenberg model systems using IBM’s superconducting quantum devices. The method is based on the application of a Lanczos expansion technique based on the computation of Hamiltonian moments and is found to offer better accuracy than conventional VQE for most cases considered, allowing for a simpler trial state and offsetting the effects of noise.
  • Item
    Thumbnail Image
    Matrix product states in quantum information processing
    Duan, Aochen ( 2015)
    We employ the newly developed Matrix Product State (MPS) formalism to simulate two problems in the context of quantum information processing. One is the Boson sampling problem, the other is the ground state energy density of an n-qubit Hamiltonian. We find that the MPS representation of the Boson sampling problem is inefficient due to large entan- glement as the number of photons increases. In the context of adiabatic quantum computing (AQC), MPS is used to find the first four moments of an n-qubit Hamiltonian to approximate the ground state energy density of the Hamiltonian. We show an advantage of using the first-four-moment method over the conventional adiabatic procedure. Future work around AQC using MPS is discussed.
  • Item
  • Item
    Thumbnail Image
    Distributed Matrix Product State Simulations of Large-Scale Quantum Circuits
    Dang, Aidan ( 2017)
    Before large-scale, robust quantum computers are developed, it is valuable to be able to classically simulate quantum algorithms to study their properties. To do so, we developed a numerical library for simulating quantum circuits via the matrix product state formalism on distributed memory architectures. By examining the multipartite entanglement present across Shor’s algorithm, we were able to effectively map a high-level circuit of Shor’s algorithm to the one-dimensional structure of a matrix product state, enabling us to perform a simulation of a specific 60 qubit instance in approximately 14 TB of memory: potentially the largest non-trivial quantum circuit simulation ever performed. We then applied matrix product state and matrix product density operator techniques to simulating one-dimensional circuits from Google’s quantum supremacy problem with errors and found it mostly resistant to our methods.
  • Item
    Thumbnail Image
    Background estimation studies for hadronically decaying tau leptons at the ATLAS experiment
    Zhang, Xuanhao ( 2018)
    This project aims to develop a data-driven technique for the estimation of the dominant background contribution in the inclusive search for new physics signals where equally charged lepton pairs are featured in the final state and where an hadronically decaying tau lepton can be found in a pair. The studies presented in this thesis were performed with data collected by the ATLAS experiment. A data driven technique has been developed for the abundant background of jets originated from the hadronisation of quarks or gluons which are mis-identified as hadronically decaying tau leptons. Mis-identification weighting factors have been measured for the extrapolation of this background into the signal region of the analysis and have been validated using a selection independent with respect to the the signal region. Systematic uncertainties have also been estimated. The work presented in this thesis will be incorporated in a general extrapolation technique within the ATLAS experiment aiming to be used by all ATLAS searches featuring hadronic tau decays in the final state.