Heriot-Watt University
ROS: The Research Output Service. Heriot-Watt University EdinburghNot a member yet
4689 research outputs found
Sort by
A unified model for enhancing citizen adoption of digital government services in the United Arab Emirates
Full text linkDigital government (DG) is leveraging information and communication technologies
(ICT), to improve public service delivery, increase public engagement, and provide
more transparent, accountable, and inclusive government services.
DG plays a pivotal role in ensuring governmental success, particularly in terms of
agility, resilience, and effective responses to disasters. Despite substantial budget
allocations for digitization, including in the United Arab Emirates (UAE), the review of
the literature and other secondary sources indicates that the adoption of DG services
remains low. Citizens still prefer visiting service centres rather than utilizing online
platforms, highlighting a critical gap in the understanding of DG adoption factors.
This study aims to investigate the factors influencing citizen adoption of DG services in
the UAE, to formulate the proposed Digital Government Services Acceptance Model
(DGSAM), inspired by the Unified Theory of Acceptance and Use of Technology
Model (UTAUT) as the underpinning theory, while integrating insights from other DG
acceptance models.
The proposed DGSAM incorporates seven constructs, namely Perceived Public Value
(PPV), Perceived User Experience (PUX), Social Influence (SI), Trust (T),
Transparency (TR), Engagement (E), and Intelligence (I), contributing to citizens’
Behavioural Intention (BI) to use DG services, and hence the actual Use Behaviour
(UB).
Embracing a pragmatic philosophy and employing a mixed-methods approach, the
research ontology acknowledges the complex nature of human behaviour and the
multifaceted aspects of DG service adoption. The epistemology recognizes the value of
integrating quantitative and qualitative data to gain a comprehensive understanding of
the factors influencing citizens' adoption behaviour.
This study employs a questionnaire with 401 digitally literate UAE citizens capable of
perceiving and using digital solutions. It also conducts semi-structured interviews with
twelve government officials and professional consultants actively involved in collecting public feedback for DG services. These interviewees serve as proxies for public
opinion, providing valuable insights into the public's perspectives.
Employing Partial Least Squares Structural Equation Modelling (PLS-SEM) as the
quantitative analysis technique, the findings reveal significant contributions from PPV,
PUX, TR, and I to users' intention to use DG services, with a direct relationship
observed between PPV and UB. However, SI, T, and E do not exhibit a significant
impact. Intriguingly, a supported direct relationship is observed between (BI) and actual
DG services usage (UB), emphasizing the practical importance of intention in predicting
real usage behaviour. Gender, age, and education nuances influence acceptance and
usage patterns, providing actionable insights for policymakers and practitioners.
The study concludes by introducing the Digital Government Services Acceptance
Model (DGSAM), a comprehensive framework developed to explain the factors
influencing citizen adoption of DG services in the UAE. This model serves as a
significant contribution to the field, providing a theoretical foundation for future
research and policy development. Additionally, it generates a list of recommendations
strategically designed to address challenges, capitalize on opportunities, and ultimately
enhance the adoption of DG services in the UAE. The study also opens avenues for
future exploration, encouraging deeper investigations into the complex interplay of
factors affecting public adoption of DG services within the UAE
Statistical and machine learning methods for low-photon imaging
Full text linkImage recovery from photon-starved measurements is a challenging problem that arises
in applications ranging from microscopy and medical imaging to astronomy and defence. In this thesis, we propose novel statistical and machine learning methods specialised for this important and difficult class of imaging problems.
As our first contribution, we introduce a new and highly efficient Markov chain
Monte Carlo (MCMC) methodology based on a reflected and regularised Langevin
stochastic differential equation (RSDE). This methodology can deal with challenges
that arise in low-photon imaging such as hard non-negativity constraints and exploding
gradients. We show that the introduced RSDE is well-posed and exponentially ergodic
under mild and easily verifiable conditions.
In our second contribution, we make use of advances in deep learning and propose
a new Bayesian methodology that can leverage deep generative priors. Deep generative models are accurate but tend to scale poorly to large imaging problems. To
address this limitation, we propose to embed a conditional deep generative prior with
a super-resolution architecture, which scales more robustly to large problems, within
an empirical Bayesian framework. This strategy allows scaling to large problems by
simultaneously computing the maximum marginal likelihood estimate (MMLE) of a
low-resolution version of the image of interest, and generating Monte Carlo samples
from the posterior of the high-resolution image of interest conditionally to the MMLE.
In our third contribution, we propose an inference strategy that leverages as image
prior a powerful class of generative models known as denoising diffusion models, which
we tailor to solve low-photon imaging inverse problems. Our approach is inspired by a
variable splitting algorithm known as the Half-Quadratic-Splitting algorithm and it is
highly computationally efficient and robust.
Finally, we integrate the proposed RSDE-based MCMC methodology into the PnP
framework. This is achieved by assuming that the prior model can be implicitly defined
by linking its gradient to a deep denoiser prior. The suggested approach improves on
the original MCMC methodology in estimation accuracy.
All proposed approaches are demonstrated with a range of experiments related to
image deblurring, denoising, and inpainting under observation noise processes that arise
in photon-starved scenarios such as the binomial, geometric and Poisson noise
Real-space imaging and modelling of gas-liquid surface scattering
No full textThe dynamics of OH scattering from atmospherically relevant organic liquid surfaces
have been studied. A rotationally cold pulsed OH molecular beam (Ek = 35 kJ mol−1
),
incident at either 0◦ or 45◦
to the surface, was scattered from perfluoropolyether, squalane
and squalene surfaces, and their peak scattered speeds and angular distributions were
determined. OH was detected with temporal, spatial, and internal-state resolution, using
planar laser-induced fluorescence.
The scattering experiment was extensively modelled using Monte-Carlo methods. Key
parameters of the experiment were explored, identifying how they affect the reliability
of speed and angular information extracted from scattering images. The finite width of
the molecular beam was found to have the greatest effect, significantly distorting the
data. This distortion was quantified, producing a set of correction factors to apply to
experimental data. Additional, independent correction factors were determined, which
account for the dependence of the detectivity on the speed of scattered molecules due to
flux-density effects.
OH was found to scatter predominantly at super-thermal speeds from all three surfaces.
The angular distributions revealed strongly directed scattering, favouring sub-specular
angles. Scattering mechanisms are inferred for non-reactive and reactive collisions. Importantly, OH addition across the double-bond, exclusive to squalene, exerts a large influence on the scattering dynamics. The development of the next-generation experiment is
described, whose design was influenced by the Monte-Carlo modelling. Initial molecular
beam characterisation is presented, and planned future experiments are described
Modelling and investigating nonlinear pulse propagation in nanophotonic periodically-poled waveguides
No full textThe recent advancement in fabrication techniques have allowed for periodically-poled
nanophotonic waveguides that can exhibit strong second- and third-order nonlinearities,
at significantly reduced operating pump levels. In this thesis, we study the evolution of
optical pulses in these structures using the unidirectional pulse propagation equation, and
demonstrate how the poling period can offer an additional degree of freedom to shape
the output spectra of nonlinear waveguides. Also, we introduce a novel architecture for
recurrent neural networks that can be trained to predict the spectral and temporal evolutions
of a pulse in different nonlinear waveguides. The presented model provides a generalised
approach to fast pulse propagation simulation, using a single neural network. Moreover,
the networks can also be designed to predict the real and imaginary components of the
pulse complex envelope, that allows the retrieval of the pulse phase, and the simultaneous
calculation of the spectral and temporal evolutions
Classification and detection of arc discharges in EV applications
Full text linkIn the era of rapid advancement in electric vehicles (EV), the detection of
potential arc faults within them has become a complex challenge on a global scale.
This study addresses this issue by laboratory experimentations and deep learning
techniques, resulting in a comprehensive framework for classifying and detecting
various types of arc faults from multiple testing circuitries with EV appliances.
The laboratory endeavours to move beyond conventional arc research involving
electrode discharge, opting instead for a practical car accident scenario involving
directly physical damages to vehicle cable. Experimental data are collected and
undergone via a designed data pipeline analysis, thereby facilitating feature
extraction and manual assessment. Subsequently, the experimental data is
employed for modelling and simulation, culminating in the creation of a signal
database tailored for deep learning applications.
Through multiple rounds of adjustments and refinements, all network
configurations achieve detection accuracies exceeding 90%, with one specific
Long Short-term Memory network showcasing an exceptional performance of
96.5%. However, these outcomes are developed on the mixed datasets of
laboratory and simulated work. It evaluates the possibilities of applying deep
learning techniques in the domain of arc fault detection research, but the
adaptability and reliability for applying it into practice EV appliances needs
further research to confirm
Quantitative methods in sustainable investment
No full textThis dissertation conducts a thorough investigation into the integration of quantitative
methods within sustainable investment frameworks, with a focus on strategic divestment
from carbon-intensive assets and the incorporation of Environmental, Social, and Governance considerations into asset management. Utilizing a multidisciplinary approach
that combines advanced computational models with analytical techniques, this research
delves into the complexities of financial markets to unearth insights crucial for the advancement of sustainable finance.
Central to this inquiry is a critique of traditional divestment strategies, which typically employ an instantaneous approach, and the introduction of a novel methodology
advocating for a rate-controlled divestment process. Employing Multi-period Portfolio
Optimization, this study meticulously examines the effects of divestment on portfolio
stability, carbon footprint reduction, and diversification, advocating for sophisticated
portfolio management strategies that align with sustainable investing principles and
demonstrate practical viability.
A significant portion of this research is devoted to addressing the lack of interactive,
open-source tools for divestment analysis. Through the development and utilization
of software solutions such as R Shiny, this dissertation contributes to democratizing
investment analysis, enabling a wider audience of investors to make informed, sustainable
investment decisions.
Additionally, this dissertation investigates the temporal effects of sustainability risk
factors on asset returns, integrating these into established financial models to provide a
holistic perspective on investment strategies. This detailed analysis facilitates a deeper
understanding of market complexities through a sustainability-focused lens.
Empirical case studies, encompassing analyses of the S&P 500, global ETFs, the UK’s
FTSE 100 index, and mixed pension funds in both the US and UK, highlight the practical applicability and effectiveness of the proposed divestment across a diverse range of
market contexts. Additionally, the factor model incorporating sustainability factors is
applied to assess the sustainability risk factor in Morningstar US indices. These studies
provide valuable insights into the operationalization of sustainability considerations in
investment strategies
Kerr-lens-modelocked single-diode-pumped femtosecond oscillators at GHz repetition frequencies
No full textAbstract and full text currently unavailable. Restricted access until 01.01.2028. Please refer to PDF
Enhancing energy demand forecasting and data imputation using deep learning : an integrated approach
Full text linkThis PhD thesis introduces an integrated approach that leverages deep learning
techniques to advance household electricity demand forecasting and data imputation within the UK energy sector. The research focuses on creating a novel system
incorporating state-of-the-art machine learning solutions for electricity demand processing and prediction. The study involves data collection from appropriate electricity demand datasets, conducting comprehensive exploratory data analysis to uncover
underlying patterns. A framework is established to process these datasets, encompassing data imputation, outlier handling, transformations, and feature scaling.
A novel missing value imputation model is developed, employing a Transformer neural network and a K-means clustering algorithm to address missing data effectively.
Subsequently, a forecasting framework for short-term residential load prediction is
presented. This modelling framework integrates a Bayesian optimisation strategy,
feature decomposition techniques, feature engineering, and percentile-based bias correction algorithms with a CNN-LSTM network to enhance prediction accuracy.
The research contributes significantly to the field of household electricity demand
forecasting and data imputation by offering a scalable and transferable framework.
The application of these methodologies yields valuable insights, not only for the
UK energy sector but also for broader applications, enabling precise predictions
and efficient demand data processing. The findings promote energy efficiency and
sustainable energy management practices.Engineering and Physical Sciences Research Council (EPSRC) funding
Direct C-H amidations of N-heterocycles
Full text linkN-Heterocycles are well-studied structures in synthetic organic chemistry and their
prevalence in the pharmaceutical industry, for example, has led to this great interest. Therefore,
the ability to directly C-H functionalisation N-heterocycles would be of great interest to the
synthetic organic chemistry community, allowing access to valuable targets without the need
for non-productive chemical steps (such as protection/deprotection, functional group
interconversions (FGI) or oxidation manipulation chemistry). This thesis will describe the
development of two complementary C-H amidations of N-heterocycles (one thermally mediated and one light-mediated) which will allow for the C-H amidation of phenanthrolines,
purines and 1,3-azoles. The term “amidation” is predominately used in this thesis to refer to
carbamoylation. Please note that the more specialist term “carbamoylation” was initially used
in Chapters 1 and 2 (and the publications contained within them). Recently, however, the more
general and well-known term “amidation” tended to be the preferred terminology to refer to
carbamoylation, and this change in notation is reflected in the publications contained within
Chapters 3 and 4.* Strictly speaking, there are two possible types of amidation:
carboxyamidation (bond-forming at the C of the amide) and N-amidation.) Amidation in this
thesis refers to carboxyamidation throughout, unless otherwise stated.
Chapter 1 will provide an introduction to the work discussed within this thesis. More
specifically, chapter 1 will introduce the Minisci reaction, a powerful method for C-H
functionalisation of electron-deficient N-heterocycles.
Chapter 2 presents a metal- and light-free C-H diamidation of 1,10-phenanthrolines.
This is the first C-H diamidation of 1,10-phenanthrolines which is capable of directly installing
primary, secondary and tertiary amides. This method is cheap, operationally simple and
scalable. Moreover, the facile 2-step (vs. previous 11-step) synthesis of a diamidated 1,10-
phenanthroline target is demonstrated.
Chapter 3 showcases the development of a direct C-H amidation of purines. As well as
being able to install primary, secondary and tertiary amides, this method is the first to
demonstrate a direct C-H amidation on a wide range of purines (e.g. xanthines, guanines,
adenines, including guanosine- and adenosine-type nucleosides). This procedure is not only cheap, operationally simple and scalable as before, but is also applicable to the late-stage
functionalisation of many biologically important molecules.
Chapter 4 communicates the development of two complementary methods (one
thermally-mediated and one light-mediated) for the direct C-H amidation of 1,3-azoles. This
work is applicable to the four most important 1,3-azoles in medicinal chemistry:
benzothiazoles, thiazoles, benzimidazoles and for the first time, imidazoles. As with the
previous chapters, the methods developed are cheap, operationally simple and scalable but also
allow for the first late-stage C-H functionalisations of 1,3-azoles. The light-mediated method
presented is the first photosensitiser-free direct Minisci-type amidation which proceeds via an
EDA. Furthermore, this newly developed photosensitiser-free Minisci-type amidation is not
only limited to azoles but can be applied to other N-heterocycles.
Chapter 5 will draw an overall conclusion of the work presented within this thesis and
discuss the possible future work.
* Since this thesis is in part by publication, it was decided to keep the terminology used within
the respective publications included in this thesis.Engineering and Physical Sciences Research Council (EPSRC
Development of a pore-scale lattice Boltzmann model for interactions between multiphase fluid and solid through wetting and reactive conditions
Full text linkStudying the subsurface flow in geo-formations requires an extensive knowledge of
multiphase flow and reactive transport. Studying these phenomena at pore scale offer
opportunities to reveal new characteristics that may be neglected when considered at large
scales. This study focuses on developing a computational model that can capture these
characteristics at pore scale. To be able to accomplish this, a lattice Boltzmann model
capable of simulating binary and ternary systems has been developed.
The model benefits from an improved wetting condition in the framework of phase-field
model applied on both ternary and binary systems when interacting with solids with a
complex topology. To achieve this, two schemes of round-off and interpolation to
implement the wetting condition are proposed. They describe how the information of
neighbouring nodes given the direction of normal vector on the solid boundary nodes
should be used. It is proven that the results of the interpolation method exhibit a good
agreement with the analytical solution as opposed to the round-off when tested in the case
of static contact angle on a circular surface for both binary and ternary systems. For the
binary system, surface-energy and geometric wetting conditions are adopted, while for
the ternary system, surface-energy model with two discretization schemes resulting in
explicit and implicit wetting conditions are developed and examined. It is proven that the
proposed implicit wetting condition enhances the accuracy of prescribed contact angles.
Furthermore, to examine the improved wetting conditions in a dynamic case,
displacement of compound droplet inside porous medium with complex structure is
investigated. It is revealed that the way the penetration is influenced by the variation of
surface tension, wettability, and density ratio completely depends on the value of Bond
number and the capillary or gravitational-dominant regimes. Furthermore, it is depicted
that in the capillary-dominant regime, the migration pattern of one droplet cannot be
evaluated independent of the other, since the factors affecting the capillary pressure and
viscous coupling on one, affects the penetration and spreading of the other. It is also
identified that ternary penetration is slower than binary with the presence of the other
droplet confining the early stage of spreading in the ternary system.
The multiphase model is further extended to be coupled with a solute transport model
which includes a first-order kinetic reactive condition resulting in a heterogeneous
reaction of fluid with solid inducing the dissolution of the solid. Different aspects of solute
transport model coupled with the reactive boundary condition are validated in isolation.
The absent features are gradually added showing the incremental steps of constructing
the final model. While the topology change mechanism is absent, the model is firstly
verified in the case of pure diffusion in an open channel. The advection term is then added
to explore the steady-state reaction of a circular CaCO₃ crystal with H+.The topology
change mechanism is then verified in the case of a fractured medium dissolution. To
couple the transport model with the improved multiphase model and simulate the reaction
of HCl and CaCO₃ (Calcite) by considering the effects of multiphase flow and CO₂ as a
product of reaction, a mass transfer term triggered by the concentration difference is
proposed. This allows for the expansion of newly emerged bubbles according to dissolved
mass of Calcite. The proposed term is tested for the case of a free bubble in an acidic
domain and satisfies the Laplace’s law instantaneously as the bubble is expanding. In
addition, details of handling the boundary conditions on the liquid-gas interface in a two-phase system where the solute only transports in the liquid phase is provided. The effects
of this are shown in the case of an acidic droplet reaching to equilibrium contact angle on
a reactive surface. The final model is then applied to simulate the reaction of HCl with an
octagonal Calcite crystal in a channel. The results are found to be in good agreement with
the experiment as well as simulation by micro-continuum DBS model in the literature. It
is shown that the presence of a non-reactive gas phase significantly influences the
dissolution. When Calcite crystals are covered by CO₂, its normalized mass is higher,
indicating reduced dissolution. Additionally, it is shown that between HCl of 0.5 wt%
and 1.0 wt%, the higher acid concentration increases bubbles residency time on the
Calcite crystal, limiting the overall dissolution. However, this cannot be generalized,
since the dissolution patterns and formation of CO₂ is affected by the combination of
reaction rate, acid’s concentration, pressure difference across the channel, and the
location of previously formed bubble. Thereby, unlike the concentration of 1.0 wt%, the
merged bubbles in other concentrations might not reach to a balance with the pressure
difference across the channel and lead to a long residency time.
The major contributions of this research can be summarised in proposing the interpolation
method for applying phase-field based wetting conditions on complex geometries, the
implicit scheme in discretizing the surface-energy wetting condition for ternary systems,
and the mass transfer term appearing in the liquid-gas interface-capturing equation
corresponding to the expansion of CO₂ as CaCO₃ undergoes dissolution
Access Repository Dashboard
Do you manage ROS: The Research Output Service. Heriot-Watt University Edinburgh ? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!