Abstract
Quantum technologies exploit quantum mechanical processes to achieve outcomes beyond the of classical machinery. One of their most promising applications is quantum simulation, particles, atoms and molecules can be examined thoroughly for the first time, having beyond the scope of even the most powerful supercomputers.Models have been useful tools in understanding physical systems: these are mathematical encoding physical interactions, which allow us to predict how the system will behave various conditions. Models of quantum systems are particularly difficult to design and test, owing to the huge computational resources required to represent them accurately. In this thesis we introduce and develop an algorithm to characterise quantum systems efficiently, by
inferring a model consistent with their observed dynamics. The Quantum Model Learning Agent is an extensible framework which permits the study of any quantum system of interest, by combining quantum simulation with state of the art machine learning. QMLA iteratively proposes candidate models and trains them against the target system, finally declaring a single model as the best representation for the system of interest.
We describe QMLA and its implementation through open source software, before testing it under a series of physical scenarios. First, we consider idealised theoretical systems in simulation, verifying the core principles of QMLA. Next, we incorporate strategies for generating candidate
models by exploiting the information QMLA has gathered to date; by incorporating a genetic algorithm within QMLA, we explore vast spaces of valid candidate models, with QMLA reliably identifying the precise target model. Finally, we apply QMLA to realistic quantum systems,
including operating on experimental data measured from an electron spin in a nitrogen vacancy centre.
QMLA is shown to be effective in all cases studied in this thesis; however, of greater interest is the platform it provides for examining quantum systems. QMLA can aid engineers in configuring experimental setups, facilitate calibration of near term quantum devices, and ultimately enable
complete characterisation of natural quantum structures. This thesis marks the beginning of a new line of research, into automating the understanding of quantum mechanical systems.
| Date of Award | 24 Jun 2021 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Anthony Laing (Supervisor) |