42337 research outputs found
Sort by
Comet taxonomies: composition-based classifications and a search for comets in the Main Belt
No full textComets are icy small bodies assumed to have remained mostly unaltered since their formation, making them key tracers of the early stages of the Solar System. While comets show a great diversity of dynamical, physical and chemical properties, efforts have been deployed to establish classifications based on these properties, with the aim of identifying different formation and evolution histories. On the one hand, from a dynamical standpoint, it has been found that comets exist in two main reservoirs before being deflected towards the inner Solar System: the Oort Cloud and the Kuiper Belt. However, a third reservoir has recently been identified as some comets have been found in the Main Asteroid Belt. Blurring the traditional divide between asteroids and comets, too few of these objects are known to understand their origin and properties. On the other hand, by quantifying the composition of the gas produced by comets, it has also been shown that classes could be established based on a high or low C₂-to-CN abundance ratio. However, the lack of a clear correlation between carbon-chain depletion and dynamical origins make this divide puzzling. Moreover, previous authors report a decrease of the measured C₂/CN ratio with the heliocentric distance of comets at the time of observation, suggesting that our understanding of C₂ production in comae might be incomplete and that C₂ based taxonomies could be biased. Since these studies typically cover short heliocentric distances (<2au) and different authors do not use consistent modelling parameters (in particular photodissociation scalelengths) to derive these abundance ratios, it is difficult to compare their findings and assess these effects. This thesis looks to bring new insights into these two challenges to established comet classifications.
First, I present a survey of comet volatiles using optical long-slit spectroscopy, aiming to investigate trends and biases in observed compositions. Spectra were acquired for 35 comets using the Isaac Newton Telescope’s Intermediate Dispersion Spectrograph. Having produced a semi-automated pipeline to reduce and analyse this large volume of data, I calculated production rates and upper limits for the main volatile species visible in the near-UV/optical range: OH, NH, C₂, CN, C₃, CH. I present a more focussed analysis of a few targets of interest such as C/2023 H2 (volatile rich) or 12P (outbursting), as well as ensemble results from the study. From these production rates, derived using a Haser outgassing model and commonly used photodissociation scalelengths, I find C₂/CN ratios consistent with a decreasing trend up to 3.5au, making most comets that were observed beyond 2au fall below the depletion threshold. I show that a correlation with perihelion distance is also possible, although I cannot clearly disentangle these two factors. When possible, I also determine and model the spatial distributions of volatiles as seen along the slit and show that a Haser model using literature scalelengths often does not reproduce the measured C₂ profiles, while CN and C₃ show a better agreement between models and observations. Using adjusted scalelengths yields larger C₂ abundances than using literature values, although it could not be determined whether this eliminated heliocentric trends.
Additionally, this thesis presents an imaging survey searching for activity in targeted Main Belt Asteroids in the hope of finding more Main Belt Comets. Using the Isaac Newton Telescope’s Wide Field Camera, r-band observations of 534 asteroids were conducted. These targets were selected based on their closeness to perihelion at the time, and on a hypothesis from previous authors that Main Belt Comets would more likely be found among objects with a longitude of perihelion close to that of Jupiter. After applying wedge photometry and point-spread function analysis methods to detect activity features via an automated pipeline, I made a candidate tail detection on images of asteroid 2001 NL19 (279870). Follow-up observations were conducted with the Liverpool Telescope at the asteroid’s following perihelion but I did not detect recurring activity, implying that the activity of this objects might not be cometary
Hybrid ceramic manufacturing for ultra-high temperature applications: investigating a guided extrusion technique for complex honeycomb structures
No full textIncreasing the efficiency of thermal processes often requires operating at elevated temperatures
towards the ultra-high temperature (∼ 1800 K) range, particularly for applications in thermal storage
systems, extractive metallurgy, solar receivers, and chemical processing. Ceramics perform exceptionally
well at these elevated temperatures; however, manufacturing ceramic structures that meet the functional
requirements of large-scale ultra-high temperature thermal processes remains challenging.
This thesis aims to establish a foundation for a hybrid ceramic manufacturing method that combines
the scalability of traditional manufacturing with advanced ceramic additive manufacturing techniques.
Focusing on ceramic honeycomb structures – widely used in thermal storage systems, heat exchangers,
and as insulation structures – this research introduces and develops a Guided Extrusion concept. This
method integrates conventional extrusion with additive manufacturing to extrude honeycomb crosssectioned
walls, enabling the creation of geometrically complex structures through layered extrusion.
The study begins with a comprehensive survey of ceramic manufacturing techniques and identifies
critical challenges in geometry, materials, and shape retention for ultra-high temperature applications.
An analytical investigation into the geometric limitations of Guided Extrusion reveals that, while the
method can partially fabricate complex structures, integrating it with existing additive manufacturing
techniques is necessary to complete the designs. Equations for rectangular extrusion profiles and
correlation figures for spherical extrusion profiles are provided to quantify these limitations.
Further, the research examines the behaviour of ceramic materials during extrusion, focusing on the
failure modes of as-extruded honeycomb columns and beams under the loading conditions imposed
by Guided Extrusion. The analysis demonstrates that conventional honeycomb profiles are prone to
buckling when attempting to create tall structures using Guided Extrusion alone. Significant interactions
between material properties, geometric parameters, and failure modes are identified, highlighting the
need for optimised profile and material combinations. To address the limitations of conventional honeycomb
profiles and paste combinations in large-scale Guided Extrusion, this thesis presents a numerical
study on an integrated drying process. This process employs hot air to uniformly dry the honeycomb
structures internally during extrusion, thereby inducing solid-like behaviour. This enhancement allows
for the stacking of more extrusion layers, facilitating the creation of complex, large-scale structures.
The study demonstrates that the degree of drying necessary to achieve the desired solid-like state can
be achieved at practical extrusion rates across a range of ceramic honeycomb profiles and materials.
Additionally, the investigation proposes a method to seamlessly integrate the drying process with the
extrusion die and examines the effects of extrusion rate and convective heating parameters. By enabling
self-supporting structures, this drying technique permits subsequent conventional thermal processing,
while potentially offering significant energy savings by utilising the honeycomb structure to uniformly
dry the extruded body.
This research advances ceramic manufacturing technology, introducing a method which can enable
the production of complex, large-scale ceramic structures suitable for ultra-high temperature applications.
By enhancing the scalability and efficiency of ceramic additive manufacturing, the Guided Extrusion
method contributes to the development of more efficient thermal systems, promoting advancements in
renewable energy, industrial processing, and sustainable technologies
Intersectionality, NGOs and executives: who has which minister’s ear?
No full textnew open access article published in
the academic journal West European
Politics reveals striking intersectional
inequalities in who gets access to
government ministers.
Government ministers have a unique ability to
further the interests of marginalised groups
through their control of policy priorities and
legislative agendas. In the current context, it’s
more important than ever that equality
organisations be able to access ministers.
However, some equality organisations are
granted this access much more often than
others. This briefing summarises what we found
about unequal access to ministers and outlines
some of the potential uses of the findings for
equalities organisation
Sport, media and national identity: the case of athletes transferring national allegiance from/to mainland China
No full textThe aim of this PhD thesis is to delineate the connections between sport and national
identity by examining how national identity is constructed in the discourse of both mass media
and social media in the context of elite athletes transferring their national allegiances from/to
mainland China. Although existing research has partially revealed the migratory patterns of
such athletes and the Chinese public’s attitudes towards these athletes, it is still not known how
these athletes are represented by Chinese mainstream mass media and how they represent
themselves online through their social media practices. Thus, this thesis was exploratory,
interpretative, and critical in nature. Qualitative content analysis and critical discourse analysis
were adopted as the main research methods of this thesis. The findings have shown that, during
the Olympic Games, the Chinese mainstream press has utilised several discursive strategies to
link the athletes who used to be Chinese citizens and the athletes who have naturalised as new
Chinese citizens to China. Furthermore, both former Chinese and new Chinese athletes have
cautiously balanced their dual national identities in their Chinese social media practices. The
significance of this PhD thesis is that it provides a new sociological understanding of the
meanings and norms of national identity during five different Olympic Games (Beijing 2008,
London 2012, Rio 2016, Tokyo 2020 and Beijing 2022) and further contributes to wider sociocultural
debates on national identity and nationalism, in particular around notions of
‘Chineseness’
Advancements in modelling the interstellar medium within cosmological simulations
Get PDFGalactic formation and evolution is driven through many high-complexity physical
processes operating over vastly differing scales, from chemical reactions to galactic
collisions. In this thesis I study the physics of the interstellar medium (ISM), which
is crucial for the evolution of stellar populations, and thus the galactic environment as
a whole. As the star formation rate is dependent upon many (often co-dependent)
properties of the ISM such as the: chemical composition, thermodynamic state,
dust population and incident radiation fields, it is incredibly challenging to describe
analytically. Therefore, the star formation rate within cosmological simulations is
modelled to reproduce the Kennicutt-Schmidt relation, an empirical law.
However, with recent simulations such as SIMBA finding success implementing H₂-
driven star formation schemes, the accuracy with which we model molecular hydrogen
populations is becoming increasingly important. As the formation and destruction of H₂
occurs on physical scales far below the resolution limit of cosmological simulations,
sub-grid models have been previously implemented to estimate molecular hydrogen
fractions.
To investigate the physical mechanisms governing the evolution of the ISM, in addition
to their impact on star formation, I integrate a novel, two-phase ISM model into the
cosmological-scale simulation suite simba. This model: explicitly tracks the co-evolution
of dust and molecular hydrogen using GRACKLE; estimates the interstellar
radiation field (ISRF) responsible for dust heating, and implements a two-phase sub-grid
model to drive star formation.
Using this model, I run two cosmological-scale simulations down to ∼ 6 in order
to investigate galactic properties. I find a large population of dust at temperatures
of ∼ 126 K (much hotter than previously assumed) and confirm that this is a result
of the ISRF dominating over all other heating terms. Running my model down to
∼ 1, I verify its predictions through comparison with contemporary observational
studies. Whilst a large population of high-temperature dust still remains at this redshift,
there exists a growing population of dust at lower temperatures, better aligning with
current expectations. Given my model’s physically motivated and self-consistent
approach, I conclude that future cosmological-scale simulations will benefit from
its implementation. With many avenues for further development — such as the
integration of sophisticated ISRF models and grain size distributions — I believe that
my work provides a solid foundation for ISM modelling in next-generation cosmological
simulations.
The explicit, self-consistent modelling of dust and H₂ introduced by my model comes
at the cost of computational performance. To understand how this is best mitigated, I
implement and benchmark GPU-accelerated versions of selected grackle functions,
with the aims of: testing the performance of simple grackle calculations to determine
the validity of a full GPU port, and developing a stratagem for such a port, whilst
retaining all current CPU functionality. I find that the use of GPU hardware will
most definitely increase grackle’s performance, allowing for calculations of higher
computational complexity whilst minimising subsequent performance costs. With the
next generation of astrophysical simulations fast approaching — alongside growing
adoption of GPU-accelerated methods within scientific contexts — I believe that a
complete GPU port of grackle is essential in securing its relevance during the coming
years
Trends in Civic Space and Elections in South Sudan: Findings from the 2025 Public Perceptions of Peace Survey
Get PDFThis report draws on 2025 polling from 4,582 respondents across South Sudan, contributing to a dataset of more than 22,000 people since 2021. It assesses public views on peace, safety, elections, civic space, and trust in political institutions. Perceptions of safety have worsened sharply: nearly one-quarter of respondents say they feel unsafe, over twice the 2024 level. This decline reflects escalating insecurity, renewed clashes between government and opposition forces, community-level tensions, and growing regional strains. Confidence that the country is “at peace” has fallen.
Support for elections, however, remains strong. Most respondents feel ready to vote and favour holding elections as planned in December 2026. Yet two-thirds anticipate electoral violence, and almost half fear the polls could spark a return to civil war. Nevertheless, these concerns do not diminish public demand for a democratic transition.
Civic space is highly constrained. Many feel unsafe discussing politically sensitive issues, and trust in political actors is low. Awareness of the Tumaini peace process is limited, especially among women. While the SPLM retains a lead in public support, disillusionment with all parties is evident. The findings indicate South Sudan is entering a high-stakes phase and call for renewed efforts to safeguard civic space, strengthen civil society, and ensure a credible electoral process
Long-horizon finite control-set model predictive control for power-electronic converters
No full textLong-horizon Finite Control Set Model Predictive Control is widely recognized for delivering superior control performance in power electronic converters when compared to conventional control methods. It offers benefits such as enhanced stability, reduced harmonic distortion, and lower switching frequencies. However, its practical implementation is significantly limited by high computational demands and hardware resource constraints, particularly in complex converter topologies. This computational burden poses significant difficulties for practical implementation on resource-constrained platforms such as FPGAs and DSPs, especially in multi-level or hybrid converter topologies.
To overcome these challenges, several algorithmic acceleration techniques have been developed, including sphere decoding, branch-and-bound pruning, and parallel computation architectures. These methods have improved execution efficiency but are still fundamentally based on search processes, limiting scalability and adaptability.
Recent developments in AI-based control through machine learning have introduced a different solution. By learning the expert control policy or function offline, these methods can emulate long-horizon predictive control behavior with drastically reduced real-time computation and hardware resource requirements.
Within this context, this research situates itself at the design and optimization of model predictive control and data-driven control. The objective is to enable scalable, hardware-efficient, and intelligent real-time control for advanced converter topologies, bridging the gap between classical MPC-based and AI-driven control architectures.
The first part of this thesis is also the first piece of work done in the PhD, building the foundation for later research. FCS-MPC was used for the converter, called the Parallel Hybrid Converter, which helped identify potential for improvements to FCS-MPC to be developed. Therefore, this part focuses on the design and optimization of PHC topology, and also the experiment environment setup. The Parallel Hybrid Converter combines two two-level converters, one utilizing low-frequency IGBT bridges and the other employing high-frequency SiC MOSFET bridges. This topology uses the SiC MOSFET bridges to offer greater dynamic control, improved harmonic performance, faster response to grid fluctuations, and reduced filtering requirements. The IGBT bridges are used to conduct the majority of the load current, while switching at a low frequency to keep losses to a minimum. To experimentally validate the proposed methodologies, a Lab-Scale Parallel Hybrid Converter Demonstrator is developed, serving as a platform for testing and verification throughout the research. The experiment results show Parallel Hybrid Converter and MPC control method achieve IGBT bridge frequencies below 1.2 kHz and grid-current total harmonic distortion below 3%, HF bridge current limited to 0.21–0.36 pu of peak phase current, and an approximate 25% overall power-semiconductor loss reduction at 1 p.u. power compared to a benchmark IGBT module converter.
The second part of this thesis focuses on improving the computational efficiency of long-horizon Finite Control Set Model Predictive Control. This thesis proposes two methods: a threaded parallel search implementation and an AI-based implementation, both of which are suitable for implementation on an embedded control platform such as an FPGA.
The threaded parallel searching method is introduced to reduce the computational limitations and challenges in Long-horizon FCS-MPC. Compared to the conventional parallel implementation method, the proposed method uses a management system that dynamically reallocates idle threads, ensuring maximum FPGA resource utilization by redirecting them to active search branches. This method significantly enhances search efficiency and hardware utilization, ensuring that all FPGA computational resources contribute to the optimization process. Compared to the conventional parallel searching algorithm, the proposed algorithm improved average calculation-time at horizon lengths of 7 by 40-45% when using a general cost-function implementation, and by 20-25% when using a sphere decoding algorithm approach.
The later chapters focus on AI-based implementation of controllers, where an offline trained 1D CNN controller based on FCS-MPC training data is designed and developed to reduce the inherent computation problem in FCS-MPC and hardware resource utilization in the FPGA. The proposed AI-based model aims to reduce the computational constraints of the conventional MPC-based methods. The results show that the proposed compressed CNN model can reduce FPGA DSP resource requirements by 75% compared to an FCS-MPC model and an MLP model, while also achieving comparable or superior performance.
These approaches are both experimentally evaluated on the Parallel Hybrid Converter, comparing their performance against conventional searching methods. The results demonstrate that both proposed techniques achieve superior control performance and reduce hardware resource requirements, highlighting their potential for practical deployment in high-performance power converters.
In the conclusion of this thesis, the main contributions of the body of work are summarized, opportunities for future improvements are discussed, and some remaining challenges are presented
Characterising small exoplanets
No full textIt was only thirty years ago that the first extrasolar planet, or exoplanet, orbiting a Sun-like star was discovered. Since then (as of October 2025), 6,022 have been
confirmed across 4,490 planetary systems, 1,013 of which host multiple planets. Whilst these exoplanets have been discovered through a range of methods, transit
photometry and radial velocity measurements have proven the most effective, accounting for∼96% of confirmed exoplanet discoveries. Through these two techniques, planetary radius and mass can be constrained to high precision. From these two parameters, planet density can be derived, enabling estimates of both atmospheric and internal compositions. Characterising small (<4 R⊕) exoplanets in this way is crucial for inferring the frequency of true Earth-analogues and assessing the uniqueness of our own planet.
However, there are several compositional trends for small exoplanets that remain poorly understood. The first is the ‘radius valley’ that separates super-Earths and sub-Neptunes, which has been consistently observed from ∼1.5–2 R⊕, and is largely without planets. Debate currently surrounds the origin of this gap, with proposed scenarios including core-powered mass-loss, photoevaporation, or that these planets are primordially rocky. Interpretations differ on the physical mechanism of atmospheric mass-loss, but the result is the same – primordially accreted atmospheres are removed in such a way that different planets are affected in different ways over different timescales, resulting in a ‘valley’ that separates a population of stripped-core planets (super-Earths) from those that have retained their H/He envelopes (sub-Neptunes). Secondly, the internal structure of sub- Neptunes is not just limited to that of a rocky core surrounded by a gaseous atmosphere, it has been theorised that these planets might hold significant fractions of ices or liquid water. It has been suggested that the radii of planets hotter than 900 K and with masses below 20 M⊕ can be reproduced assuming ice-dominated compositions without significant gaseous envelopes. However, it has also been argued that the existence of small planets with hydrogen atmospheres is consistent with the data, once thermal evolution and mass-loss are properly accounted for. This means that there is a strong degeneracy between water-world and silicate/iron-hydrogen models, and that the characterisation of larger sub-Neptunes in this region of the mass–radius diagram can be used to determine planetary evolution and formation pathways.
With our understanding still limited regarding the origins of these compositional trends, taking steps towards improving characterisation methods of bodies and systems in this size range is vital. Improving our understanding of the origins of the radius valley and the diverse pathways of planetary development will finally help us to ascertain the uniqueness of our own solar system and planets, which is a question that humanity has attempted to answer since the beginning of time
Estimating genetic and environmental sources of variance for depression
Get PDFMajor depressive disorder (MDD) is a highly prevalent psychiatric disorder that is now the leading cause of worldwide disability in terms of years lived with disability. In the majority of Western countries, the lifetime prevalence of MDD typically varies between 8% and 12%. There are consistently established relationships with female gender, alcohol abuse, diabetes, and poor social relationships. The high prevalence and disability associated with MDD make research aimed at understanding its aetiology and developing effective treatments a priority.
MDD aggregates within families and the heritability of MDD has been estimated as 37% (SE 5%) in a meta-analysis of twin studies and 32% (SE 9%) using genomic similarity among unrelated individuals. Given the genetic contribution to MDD, genetic studies are a potential means of understanding its aetiology as well as identifying new drug targets. Despite this substantial genetic contribution to its aetiology, candidate gene and genome-wide association studies, including a mega-analysis of more than 20,000 individuals with 9240 cases and 9519 controls in the discovery sample, have failed to identify significantly associated specific genetic variants. Nonetheless, genome-wide association and related studies have shown that MDD is a genetically complex disorder where risk is proposed to result from the cumulative effects of many low-penetrance genetic variants.
Increasingly it is also recognised that a diagnosis of MDD may group together individuals who suffer from causally distinct conditions. Some studies indicate that the heritability estimates of MDD differ by sex with female MDD showing higher heritability than male MDD suggesting that the genetic causes may be somewhat distinct. Further, it has been suggested that both age of onset and single versus recurrent episode illness course may have somewhat differing genetic aetiologies. These findings highlight the substantial heterogeneity of MDD, which may further impede the search for genetic causes.
There is therefore an urgent need to increase sample sizes and to refine and stratify the phenotype to reduce heterogeneity of phenotypic measurements, and measurement error of MDD with the aim of identifying more genetically homogenous targets for better powered association studies. Pedigree-based genetic studies are an efficient means for dissecting trait heterogeneity because they are able to capture all additive heritability whilst matching for key confounds present in studies of unaffected subjects
Computational modelling of behavioural differences in anxiety disorders
Get PDFAnxiety is one of the most common types of mental disorders which manifests itself in different forms and intensities based on the types of situations inducing it and which can cause significant distress in people experiencing it. Despite the high prevalence of anxiety in the general population, a mechanistic understanding of anxiety disorders is still limited, with difficulties in providing precise diagnosis and predicting treatment outcomes due to its varied nature and high comorbidity with other disorders. In recent years, computational modelling has been increasingly used to formalise and investigate the mechanisms behind anxiety and other psychiatric disorders in an effort to provide precise, measurable descriptions of the key processes involved, ranging from the neural to the behavioural level. In this thesis, we investigate three areas of human behaviour known to be affected by anxiety: aversive learning, economic decision-making and estimation of uncertainty.
The first study of this thesis investigates whether anxiety is linked to overgeneralisation of threat learning by modelling data collected in an avoidance learning task using Reinforcement Learning and Drift Diffusion models. A first experiment, previously conducted in the Seriès' Lab, found that anxious individuals overgeneralise threat learning due to impairments in discriminating between safe and aversive stimuli, poorer instrumental performance, higher perceived difficulty and diminished punishment sensitivity. In this thesis, we replicate this analysis on a second dataset collected using the same task on clinical participants diagnosed with Generalised Anxiety Disorder (GAD). We found that the original results did not replicate in this new dataset due to methodological differences with the initial design.
The second study presented in this thesis addresses the impact of anxiety on individual propensities to risk and loss aversion in a gambling task delivered online. The experiment also investigates how individual levels of risk and loss aversion affected individual attitudes towards vaccines and decisions taken during the COVID pandemic, and whether these decisions were modulated by anxiety levels. Using a hierarchical Bayesian version of Prospect Theory models, we found increased loss aversion in individuals with higher levels of GAD and no differences in risk aversion. We also found no links between individual levels of risk and loss aversion and attitudes towards vaccines and COVID, possibly due to the economic decision-making task used in the study.
The third study in this thesis examines how anxiety may affect estimation of uncertainty in the environment. When we interact with the environment, several types of uncertainty like expected and unexpected uncertainty affect our decisions. Each of these types of uncertainty have different, sometimes opposite, effects on how we learn from the environment. First, we report the results of a systematic review of the computational anxiety literature regarding estimation of uncertainty.
We find that the literature agrees on anxiety being linked to an overly volatile view of the environment, which results in an impairment in adjusting learning as a result of changes in the environment. We also highlight current challenges and limitation due to the heterogeneity of tasks and computational models used in the literature, and we present a case study in which we compare two widely used families of computational models used in the literature, namely Reinforcement Learning models and the Hierarchical Gaussian Filter. Second, we design a set of experiments aimed at testing whether different uncertainty manipulations and the valence of the outcomes play a role in uncertainty processing and anxiety disorders. We find that, when moving from a more stable to a more volatile environment, individuals with high levels of anxiety, depression, PTSD and autistic traits show reduced adaptation of learning rates in punishing environments and increased adaptation of learning rates and greater reliance on past choices in rewarding environments. We find that intolerance of uncertainty mediates the effects of anxiety, depression, PTSD symptoms and autistic traits on punishment learning rate adaptation, while intolerance of uncertainty only mediates the effects of state anxiety, depression and PTSD symptoms on reward learning rate adaptation. We propose intolerance of uncertainty as a transdiagnostic factor that may capture differences in uncertainty processing across multiple disorders. We also discuss differences in results obtained through Reinforcement Learning models and Hierarchical Gaussian Filters and we discuss methodological challenges in using these models within a hierarchical Bayesian framework
Access Repository Dashboard
Do you manage Edinburgh Research Archive? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!