17837 research outputs found
Sort by
Predicting London’s Precipitation: A Spatio-Temporal Neural Network Approach
No full textThis study presents a data-driven approach to forecasting total precipitation in London using an Artificial Neural Network (ANN) within a spatio-temporal framework. Leveraging ERA5 data from 2010 to 2025, the methodology includes automated NetCDF extraction, feature engineering with lagged precipitation and cyclic time encodings, and dimensionality reduction via a trained Autoencoder. The ANN, designed in a GenCast-style architecture, was trained using the Adam optimiser over 50 epochs and achieved strong performance. SHAP analysis highlighted the importance of lag features and seasonal time variables, enhancing interpretability and supporting the model’s application in urban flood risk management and climate resilience
Rich picture - A history of Sheerwater Housing Estate
No full textGraphical representation of the history of Sheerwater Housing Estate as perceived by long term resident
French Nineteenth-Century Art Writing as Audio Description: the case of Edouard Manet
Get PDFThis chapter compares the ‘traditional’ audio description of Edouard Manet’s 1863 masterpiece Olympia with descriptions of the painting by 19th-century French critics made when it was first put on public display in Paris in 1865. This comparison suggests that descriptions that include references to artistic techniques, personal opinion, and the various ways a beholder looks at and responds to a work of art produce a more engaging and evocative audio description than the supposedly objective and neutral texts that are recommended by best practice guides. By embracing the plurality of responses to a painting, and acknowledging that different people view paintings in different ways, this chapter advocates for a more creative approach to audio description that might better capture the experience of being moved by a work of art
OSTRICH2: Solver for Complex String Constraints
Get PDFWe present OSTRICH2, the latest evolution of the SMT solver OSTRICH for string constraints. OSTRICH2 supports a wide range of complex functions on strings and provides completeness guarantees for a substantial fragment of string constraints, including the straight-line fragment and the chain-free fragment. OSTRICH2 provides full support for the SMT-LIB theory of Unicode strings, extending the standard with several unique features not found in other solvers: among others, parsing of ECMAScript regular expressions (including look‐around assertions and capture groups) and handling of user‐defined string transducers. We empirically demonstrate that OSTRICH2 is competitive to other string solvers on SMT-COMP benchmarks
AI-Enhanced Muon Shield Design for Background Mitigation in Next-Generation Intensity Frontier Experiments at CERN
Get PDFThe North Area at CERN uses a 400 GeV/c proton beam from the SPS to supply secondary beams to fixed-target experiments. Among these, K12 supports NA62, a precision experiment studying ultra-rare Kaon decays. With NA62 concluding after CERN's Run~3, the beamline to Experimental Cavern North 3 (ECN3) will become available, offering an opportunity for new experiments -HIKE with SHADOWS and NaNu or SHiP with SND@SHiP - to search for physics beyond the Standard Model. This thesis presents studies that were essential in demonstrating the feasibility of the SHADOWS and NaNu experiments, substantially contributing to their proposal.All these experiments target the intensity frontier, requiring a significant reduction of radiation background at up to 30 times higher instantaneous proton intensities than those delivered so far. Since muons constitute the dominant source of background, mitigation strategies were extensively analysed and developed. Magnetic muon shields based on magnetised iron blocks were identified as effective muon sweepers. Comprehensive design studies were conducted bringing their design from a theoretical concept to demonstrated feasibility, addressing integration requirements and beamline-specific challenges.To further enhance the magnet design within the beamline's high-radiation environment, a novel AI-aided approach was developed. This method enables tailoring the magnet design to the specific background conditions in the cavern. The effectiveness of this method was demonstrated across multiple case studies presented in this thesis
Low Energy Nuclear Recoil Band Tuning in the LUX-ZEPLIN Experiment
Get PDFCenturies of astrophysical observations have indicated the presence of an in-visible, weakly interacting matter known as dark matter, with the most popular candidate being the Weakly Interacting Massive Particle (WIMP).This thesis focuses on calibrations aspect of the underground liquid xenonexperiment, LUX-ZEPLIN. LZ uses a dual-phase time projection chamber (TPC) to observe a WIMP-nucleon scattering signal, with its second science result released inAug. 2024. Accurate discrimination between electronic recoil (ER) and nuclear recoil (NR) signals at low energies is vital for improving dark matter detector sensitivities.The main focus of this thesis is on LZ’s lowest energy neutron source, YttriumBeryllium (YBe), a photoneutron source that was first deployed after the first science result. Obtaining a clean selection of the YBe relied on removing an unexpected population of above-anode gas events obscuring the NR region. These events occur above the anode that drift down and create light in the high-field region. This selection work resulted in obtaining a sample of 231 YBe events, fitting the 1σ and2σ contours from YBe simulations.Using this selection data, a 6-parameter fit was conducted for the NR modelin LZ called Noble Element Scintillation Technique (NEST), using Markov ChainMonte Carlo. This is a novel high dimensional fit for tuning the low energy NR region using the lowest energy calibration source in LZ. The fitting process impacted the lower mass WIMPs significantly, quantified through assessing the impact on the number of generated WIMPs from the default NEST models and the fitted models.Using three different simulated mass WIMPs (i.e. 6 GeV, 10 GeV and 100 GeV)and comparing both simulated models showed a 12% reduction in the numberof generated 6 GeV WIMP events, a 5% reduction for 10 GeV WIMP events butminimal impact on 100 GeV events.These results obtained higher uncertainties than those from the default model, leading to questions surrounding whether there were sufficient statistics to constrainthe model for a 6-dimensional fit. A parallel tuning analysis was conducted using a larger simulated YBe dataset, to investigate the uncertainties, showing a similar reduction in generated events and a smaller uncertainty to the previous tuning, yet still higher than the default NEST.Both tuning analyses impact the intended low energy and mass WIMP region most, with indication of potential overestimation from the currently used NEST model parameters. Yet the higher uncertainties may suggest that the data or modelmight not have sufficiently constrained the parameters requiring the need for further statistics and research
A Retreat from Global Solidarity:The Consequences of International Aid Cuts
Get PDFAs the US, UK and other countries slash development aid, Melita Lazell and Ivica Petrikova warn of the devastating consequences for global health, security, and diplomacy – and argue that the retreat from solidarity is a perilous gamble for us all
Firm–education–industry association linkages:Driving the territorial embeddedness of business services multinational corporations in Romania?
Get PDFThis study examines the nature and implications of linkages between multinational corporations and local institutions across peripheral regions. Analysing the development of outsourced and offshored business services in Romania, the study highlights the role of firm–education–industry association linkages in driving the territorial embeddedness of multinational corporations into host country regions. Firm–education–industry association linkages facilitated changes in higher education curricula to supply firm-specific skills, the development of advanced technical and management skills, and a programme of state policies privileging foreign capital. While this industrial and institutional transformation facilitated Romania’s move up the value chain into more advanced business services, it simultaneously drove forms of corporate capture and dependency, reproducing a flexible, co-opted workplace labour regime
Data sources:
Access Repository Dashboard
Do you manage Royal Holloway - Pure? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!