Language-based fashion image editing has promising applications in design, sustainability, and art. However, it is considered a challenging problem in computer vision and graphics. The diversity of human poses and the complexity of clothing shapes and textures make the editing problem difficult. Inspired by recent progress in editing face images through manipulating latent representations, such as StyleCLIP and HairCLIP, we apply those methods in editing the images of full-body humans in fashion datasets and evaluate their effectiveness. First, we assess different methodologies to find a latent representation of an image via Generative Adversarial Network (GAN) inversion; then, we apply three image manipulation schemes. Thus, a pre-trained e4e encoder is initially utilized for the inversion process, while the results are compared to a more accurate method, Pivotal Tuning Inversion (PTI). Next, we employ an optimization scheme that uses the Contrastive Language Image Pre-training (CLIP) model to guide the latent representation of an image in the direction of attributes described in the input text. We address the problem of the accuracy and speed of the process by incorporating a mapper network. Finally, we propose an optimized mapper called Text-Driven Garment Editing Mapper (TD-GEM) to achieve high-quality image editing in a disentangled way. Our empirical results show that the proposed method can edit fashion items for changing color and sleeve length. / Språkbaserad bildredigering inom mode har lovande tillämpningar inom design, hållbarhet och konst. Det betraktas dock som ett utmanande problem inom datorseende och grafik. Mångfalden och variationen av mänskliga poser och komplexiteten i klädform och texturer gör redigeringsproblemet svårt. Inspirerade av den senaste utvecklingen inom redigering av ansiktsbilder genom manipulation av latenta representationer, såsom StyleCLIP och HairCLIP, tillämpar vi dessa metoder för att redigera bilderna av fullständiga mänskliga kroppar i mode-dataset och utvärderar deras effektivitet. Först jämför vi olika metoder för att hitta en latent representation av en bild via så kallade Generative Adversarial Network (GAN) inversion; sedan tillämpar vi tre bildmanipulationsscheman. En förtränad (eng: pre-trained) e4e-encoder model används först för inversionsprocessen, medan resultaten jämförs med en mer exakt metod, Pivotal Tuning Inversion (PTI). Därefter använder vi en optimeringmetod som använder Contrastive Language Image Pre-training (CLIP) -modell för att vägleda den latenta representationen av en bild i riktning mot attribut som beskrivs i inmatningstexten. Vi tar upp problemet med noggrannhet och hastigheten i processen genom att integrera en mapper-nätverk. Slutligen föreslår vi en optimerad mapper som kallas TD-GEM för att uppnå högkvalitativ bildredigering på ett lösgjort sätt. Våra empiriska resultat visar att den föreslagna metoden kan redigera modeobjekt för att ändra färg och ärmens längd.

The discovery of DNA has been one of the biggest catalysts in genomic research. Sequencing has enabled us to access the wealth of information encoded in DNA and has provided the basis for ground-breaking achievements such as the first complete human genome sequence. Furthermore, it has tremendously advanced our understanding of life-threatening genetic disorders and bacterial and viral infections. With the recent advent of next generation sequencing (NGS) technologies, sequencing became accessible to the majority of researchers and made metagenomic sequencing widely available. However, to realise its true potential, sophisticated and tailor-made bioinformatic programs are essential to translate the collected data into meaningful information. My thesis explored the potential of resolving fine-scale variation in NGS data. The identification and correction of artificial fine-scale variation in the form of biases and errors is imperative in order to draw valid conclusions. Furthermore, resolving natural fine-scale variation in the form of single nucleotide polymorphisms (SNPs) and closely related species or strains is critical for the development of effective treatments and the characterisation of diseases. In recent years, Illumina has emerged as the global market leader in DNA sequencing. However, biases and errors associated with this high-throughput sequencing technology are still poorly understood which has precluded the development of effective noise removal algorithms. In addition, many programs were not designed for Illumina data or metagenomic sequencing. Therefore, a better understanding of the idiosyncrasies encountered in Illumina data is essential and programs must be tested and benchmarked on realistic and reliable in silico data sets to reveal not only their true capacities but also their limitations. I conducted the largest in vivo study of Illumina error profiles in combination with state-of-the-art library preparation methods to date. For the first time, a direct connection between experimental design factors and systematic errors was established, providing detailed insight into the nature of Illumina errors. Further, I tested various error removal techniques and developed a sophisticated Illumina amplicon noise removal algorithm, enabling researchers to choose optimal processing strategies for their particular data sets. In addition, I devised several simulation tools that accurately reflect artificial and natural fine-scale variation. This includes a flexible and efficient read simulation program which is the only program that can directly reflect the impact of experimental design factors. Furthermore, I developed a program simulating the evolution of a virus into a quasi-species. These programs formed the basis for two comprehensive benchmarking studies that revealed the capacities and limitations of viral haplotype reconstruction programs and taxonomic classification programs, respectively. My work furthers our knowledge of Illumina sequencing errors and will facilitate more accurate and effective analyses of sequencing data sets.

<p>This thesis presents and uses the techniques of computational chemistry to explore two different processes induced in human skin by ultraviolet light. The first is the transformation of urocanic acid into a immunosuppressing agent, and the other is the enzymatic action of the 8-oxoguanine glycosylase enzyme. </p><p>The photochemistry of urocanic acid is investigated by time-dependent density functional theory. Vertical absorption spectra of the molecule in different forms and environments is assigned and candidate states for the photochemistry at different wavelengths are identified. </p><p>Molecular dynamics simulations of urocanic acid in gas phase and aqueous solution reveals considerable flexibility under experimental conditions, particularly for for the <i>cis</i> isomer where competition between intra- and inter-molecular interactions increases flexibility. </p><p>A model to explain the observed gas phase photochemistry of urocanic acid is developed and it is shown that a reinterpretation in terms of a mixture between isomers significantly enhances the agreement between theory and experiment , and resolves several peculiarities in the spectrum. </p><p>A model for the photochemistry in the aqueous phase of urocanic acid is then developed, in which two excited states governs the efficiency of photoisomerization. The point of entrance into a conical intersection seam is shown to explain the wavelength dependence of photoisomerization quantum yield. </p><p>Finally some mechanistic aspects of the DNA repair enzyme 8-oxoguanine glycosylase is investigated with density functional theory. It is found that the critical amino acid of the active site can provide catalytic power in several different manners, and that a recent proposal involving a S<i>N</i>1 type of mechanism seems the most efficient one.</p>

Hydrocarbon selective catalytic reduction (HC-SCR) is emerging as one of the most practical methods for the removal of nitrogen oxides (NOx) from light-duty-diesel engine exhaust gas. In order to further promote the chemical reactions of NOx-SCR by hydrocarbons, an understanding of the HC-SCR process at the molecular level is necessary. In the present work, a novel surface-reaction mechanism for HC-SCR is set up with emphasis on microkinetic analysis aiming to investigate the chemical behaviour during the process at a molecular level via detailed elementary reaction steps. Propane (C3H8) is chosen as the reductant of HC-SCR. The simulation is designed for a single channel of a monolith, typical for automotive catalytic converters, coated with a silver alumina catalyst (Ag/Al2O3). The complicated physical and chemical details occurring in the catalytic converter are investigated by using the numerical method of computational fluid dynamics (CFD) coupled with the mechanism. The C3H8-SCR reaction mechanism consists of 94 elementary reactions, 24 gas-phase species and 24 adsorbed surface species. The mechanism is optimised by tuning some important reaction parameters against some measurable data from experiments. The optimised mechanism then is validated with another set of experimental data. The numerical simulation shows good agreements between the modelling and the experimental data. Finally, the numerical modelling also provides information that is difficult to measure for example, gas-phase concentration distribution, temperature profiles, wall temperatures and the occupation of adsorbed species on catalyst surface. Consequently, computational modelling can be used as an effective tool to design and/or optimise the catalytic exhaust aftertreatment system.

Fuels: Non-renewable fossil fuels are the largest man-made contributor to global warming. Biofuel market share will increase, promoted by renewability, inherent lower net CO\(_2\) emissions, and legislation. The environmental and human health impact of diesel exhaust emission particulate matter (PM) is a major concern. Fossil diesel PM aftertreatment systems exist. Near future fuel PM research and the evaluation of current aftertreatment technology, highlight a route for future development. Using a holistic approach this body of work studies the interdependence between the fuel, the CI engine and associated aftertreatment system. The overall objective of this thesis is the evaluation of current diesel aftertreatment using renewable near future fuels. Diesel blends with 1st and 2nd generation biodiesel fuels are viable. Carefully selected blends like B20G10 can make all round regulated emission improvements. Green additive dimethyl carbonate (DMC) enhances diesel combustion by oxygenation. Regulated emissions of THC, CO (>30% reduced) and PM (50% reduced) for 2% DMC in diesel. 1st generation biodiesel (RME) is widely available. Combustion produces significantly less solid PM than diesel (<50%) and slightly more liquid PM. RME produces less particulate at nearly all particle sizes, but more of the small nano sized liquid SOF. The potentially negative health effect of nano-sized SOF material raises questions. There is a case for more research into the health effects of nano-sized SOF material. Oxygenated fuel combustion PM contains more voids (facilitates DPF oxidation), plus unregulated carcinogenic compounds are reduced. ii Aftertreatment: The efficiency of the oxidation catalyst for the near future fuels was comparable to diesel and PM matter at all particulate sizes is reduced for all near future fuels tested. Aftertreatment total PM filtration levels are >90% by mass >98% by number, for all fuels. For synthetically produced 2nd generation gas to liquid (GTL) fuels there are potentially DPF regeneration implications. Current aftertreatment solutions are as effective for bio alternatives as they are for fossil diesel. Biodiesels, green additives and aftertreatment are effective clean emissions improvements, until the promise of true zero emission vehicles is realised.

The world is facing the challenge of global warming and environment protection. On the other hand, the demand of electricity is growing fast due to economic growth and increase in population. Since the growth in demand is also a heavy factor in energy equations, then the renewable energy alone is not able to generate enough electricity to fill the gap within a short time of period. Therefore, fossil fuel such as coal fired power plants cannot be ruled out immediately due to their generation capacity and flexibility in load following. However, any new coal fired stations should be cleaner compared with traditional power plants. Supercritical power plants are one of the most suitable choices for environmental enhancement and higher efficiency. However, there has been an issue of whether or not to adopt this technology in the UK because it is not clear whether the performance for SC plants can satisfy the British Gird Code requirement. This thesis reports a study of dynamic responses of SC power plants through mathematical modeling, and simulation for Gird Code compliance. It also presents a new control strategy based on an alternative configuration of generalized predictive control for power plant control.

Hollow low density microspheres were prepared by adsorbing kaolin nanoparticles onto a polystyrene (PS) template. A cationic polyelectrolyte, poly(diallydimethylammonium chloride), was initially adsorbed on the PS to render the surface cationic enabling the kaolin nanoparticles to form a shell structure due to electrostatic attraction forces. An increase in ionic strength by the addition of 0.1 M NaCl increased the amount of polyelectrolyte adsorbed at the PS surface. Solvent relaxation NMR experiments indicated that solvent molecules were bound at the PS surface suggesting attached polymer chains. Zeta potential experiments indicated a change in the surface potential at the PS interface due to the addition of a polyelectrolyte, as little as 0.1 g m-2 of polyelectrolyte was required to change the PS surface potential. Surface saturation of the PS resulted in a zeta potential of 63.5 \(\pm\)2.6 mV, at pH 6.6 \(\pm\)0.2. Kaolin adsorption was determined by reduced sediment volume experiments and observed by the use of scanning electron microscopy (SEM). The influence of an increase in the kaolin ratio and the calcination process on the microsphere structure was investigated. When calcined at 1000 ºC for 60 min the microsphere properties exhibited a particle size (d50) of 11.2 \(\pm\)0.4 \(\mu\)m and a bulk density of 0.13 \(\pm\)0.01 g/cm3. The internal structure of the calcined microspheres were characterised by the use of SEM and focused ion beam (FIB) instruments. The characteristic properties of the calcined kaolin microspheres were determined by thermogravimetric analysis (TGA), X-ray diffraction (XRD), X-ray fluorescence (XRF), and X-ray photoelectron spectroscopy (XPS).

Anaerobic digestion technology has not gained widespread acceptance on UK farms due mainly to the long return on investment periods involved. It has been suggested that co-digestion of agricultural and industrial wastes may enhance the economic viability of such installations. Batch and continuous digestion of cattle slurry and organic industrial wastes was carried out in specially constructed pilot plant digesters, to determine optimum mixtures of waste and digester loading rates. A total of 10 different wastes were tested, on a batch digestion basis, for their potential to co-digest with cattle slurry. Of these, 3 were chosen for continuous pilot plant trials, due to either a need to provide a disposal route for the waste, or positive effects of the waste on methane productivity. Chicken manure was found to slightly enhance methane productivity, but ammonia inhibition of methanogenic bacteria was noted over time. The organic fraction of municipal household waste (OFMSW) significantly enhanced digester methane productivity, while fish offal (FO) slightly enhanced methane productivity when added to the digester in small quantities, but quickly caused digester failure when added in larger amounts. An economic model of a digestion facility was developed and used to show the financial benefits of co-digestion.

Electrical power generation from wind energy has been recognised as one of major realistic energy sources in CO2 emission reduction worldwide. However, matching power generation with the load demand remains a great challenge, due to the nature of wind energy intermittency. The thesis addresses this issue by developing a new system with the structure of a hybrid connection of the wind turbine and compressed air energy storage. A scroll air motor is adopted to serve as an “air-electricity transformer” to compensate the power output during the period of low wind speed. The complete mathematical model and a suitable management and control strategy for the whole hybrid system are developed in the thesis. The simulation study has demonstrated that the proposed new hybrid wind turbine system is feasible and has potential for industrial applications. The prototype of hybrid wind turbine is successfully constructed in the author’s group. And the experimental results finally prove the concept of the new hybrid wind turbine. At last, the experimental system result analysis is described in this thesis. Simulation and experimental study shows that the proposed hybrid wind turbine system is technically feasible with energy efficiency around 50%.

The following research thesis focuses on methods of controlling nitrogen oxides (NO_x) and particulate matter (PM) emissions emitted from a low temperature diesel exhaust. This involves studying the influence of hydrogen (H₂) on various aftertreatment devices such as hydrocarbon selective catalytic reduction (HC-SCR) over silver-alumina (Ag-Al₂O₃) catalysts for lean NO_x reduction, platinum diesel oxidation catalysts (DOC) for nitrogen dioxide (NO₂) production and passive regeneration methods for the diesel particulate filter (DPF). H₂ was implemented on-board either through diesel exhaust gas fuel reforming or via the simulation of ammonia (NH₃) dissociation. Both methods showed to be very effective in enhancing the activity of a silver HC-SCR catalyst for the reduction of NO_x with conversions reaching 90% with the aid of an upstream DPF. A combined DOC and catalysed DPF (cDPF) configuration proved promising for passive regeneration in the presence of reformed exhaust gas recirculation (REGR). The addition of H₂ over the DOC led to an improved catalyst light-off temperature and increased rate of oxidation for NO₂ production. Implementing filtered EGR (FEGR) removes the hydrocarbon (HC) and soot recirculation penalty, thus minimising particulate growth which results in a significantly reduced engine-out soot emission during exhaust gas recirculation (EGR) and hence, an improved NO_x/soot ratio. Introducing fuel components which enhance the cetane number and oxygenate the diesel fuel allow better control of the NO_x/soot trade-off with improved soot oxidation properties.