A reconfigurable silicon quantum photonic device for machine learning applications

  • Konstantina I Koteva

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)

Abstract

Quantum machines exploiting particle entanglement and superposition are expected to achieve computational advantages beyond classical computers. Silicon photonics provides a stable and manufacturable approach to quantum information processing. While a universal linear optical quantum computer is still far from our reach, reconfigurable multi-dimensional encoding on multiple photons is scalable and provides an interesting route to noisy intermediate-scale quantum applications in photonics.

Qudits are quantum states with dimension higher than traditional qubits and offer certain advantages for quantum information processing. With their intrinsically higher information capacity, qudits have been used to obtain larger violation of local-realistic theories thus playing an important role in both quantum communication and quantum computing. A fundamental prerequisite to enable qudits as carriers of quantum information in such applications is the demonstration of high-fidelity, arbitrary operations upon qudits.

In this thesis, we introduce a silicon quantum photonic device capable of performing arbitrary unitary operations on a four-dimensional qudit controlled by a qubit encoded in an ancillary register. We tackle some of the challenges in component characterisation of quantum hardware and perform thorough testing of this device. We then study the device performance by carrying out a selection of tomographic state and process reconstructions and demonstrate the highest fidelity integrated quantum process tomography results for four dimensions reported to date. Finally, we use this device as a quantum simulator to demonstrate an adaptive learning protocol for estimating the gaps between energy levels of Hamiltonian systems and apply it to real-life examples.

These results confirm the consistently high performance of silicon quantum photonic technology, as evidenced by extensive testing and execution of protocols involving multiple photonic qudits. This work therefore constitutes a key step towards practical applications of integrated quantum photonic devices.
Date of Award25 Jan 2022
Original languageEnglish
Awarding Institution
  • The University of Bristol
SupervisorAnthony Laing (Supervisor) & Jonas H Rademacker (Supervisor)

Cite this

'