143174 research outputs found
Sort by
Do Investors care about the Rainforest? Evidence from voluntary carbon offsets around the world
No full textWe explore how investors react to firms’ biodiversity–focused activities in the Voluntary Carbon Market. The average stock price reaction of forestry carbon offsetting with and without biodiversity impact is not significantly different from zero until the end of 2022. Following a Guardian article claiming rainforest carbon offsets to be ‘worthless’ in January 2023, the announcement returns to carbon credit retirements turned significantly negative. This effect is not significant for carbon credits contributing to biodiversity conservation certified by the Climate, Community & Biodiversity CCB Standard, but it is significantly negative for other credits. Our results show that investors care about both the biodiversity impact of carbon credits as well as the climate–change mitigation integrity of offsetting activities
The effect of microsolvation on the mechanism, kinetics and energetics of a reductive amination reaction
No full textThe reductive amination of carbonyl compounds is widely employed to synthesize functionalized amines in solution. The presence of water is known to impact the reaction rate but the underlying phenomena are poorly understood. Here, we combine nuclear magnetic resonance spectroscopy and quantum mechanical (QM) calculations to elucidate the effect of microsolvation on the mechanism, kinetics and energetics of the reaction between benzaldehyde and 4-(trifluoromethyl)-aniline in tetrahydrofuran. Kinetic data with varying water content indicate an autocatalytic effect. We explain this process by including two water molecules in QM calculations, demonstrating the transition-state stabilization that results from hydrogen bond formation. Through an extensive investigation of different QM methods, we find a “high-low” computational scheme with the G2MP2 method and SMD solvation model yields good quantitative agreement with experimental evidence, providing a theoretical approach to accelerate the development of other reductive amination reactions
Vector-borne diseases and the Syrian conflict: a Systematic Review of literature from Syria and neighbouring, refugee-hosting countries
No full textObjectives: Syria’s conflict has forced more than half its populations from their homes, decimated its health system and water and sanitation infrastructure, leading to an increase in communicable diseases, particularly vector-borne diseases (VBDs.) This systematic review explores the epidemiological burden and geographical distribution of VBDs in Syria and neighbouring, refugee-hosting countries following the onset of the Syrian uprisings. Methods: Four databases were searched using appropriate MeSH terms. Included studies reported on data collected between March 2011 and May 2024 on VBD-affected populations in Syria, Turkey, Jordan, Lebanon or Iraq. Data were extracted, study quality assessed, and findings synthesized in narrative form. Results: 33 studies were included; all but one reported on leishmaniasis, the exception reported on malaria. 16 were from Turkey, nine from Lebanon, five from Syria and three from Jordan. Data showed an increase over time in the numbers of reported leishmaniasis cases and geographical spread as well as barriers to healthcare access for Syrian refugees. Conclusions: In this systematic review, Leishmania was the primary VBD described in this context. Studies from country-specific health ministries mostly reported data whereas smaller studies added more granular information including around healthcare access. Additional studies are needed to identify vector reservoir populations and to investigate the burden of other VBDs in this region
Capturing spatio-temporal variations in SDG interactions and prioritizations
No full textUnderstanding the varying degrees of synergies and trade-offs among Sustainable Development Goals (SDGs), as well as the influential goals within these interactions, is crucial for identifying transformative governance actions. However, the spatiotemporal dynamics of SDG interactions and priorities remain unclear. Here, we analyzed global and regional variations in SDG interactions, synergistic performance, and key contributing goals from 2000 to 2022 using network methodology, index analysis, and machine learning. Europe exhibited high SDG synergies, low synergy-network modularity, and strong synergistic performance, while Western Asia showed the opposite pattern. Despite overall improvement in SDG synergy proportions and synergistic performance, global progress was constrained by declining synergies related to SDG 5 (gender equality) and increasing fragmentation of the SDG synergy network. We also observed substantial spatio-temporal changes in the impact of each SDG within the interaction network and in its contribution to overall synergistic performance. Furthermore, we applied an ensemble random forest model to assess SDG mentions and co-occurrences in 1944 SDG interaction studies. SDGs 13 (climate action) and 9 (industry, innovation and infrastructure) emerged as the most and least frequently discussed goals, respectively. Interactions among seven SDGs (2, 6, 7, 8, 12, 13, 15) formed a typical SDG nexus, reflecting a critical human-nature relationship chain. Overall, to advance global SDG attainment, we emphasize the importance of accelerating progress on SDG 3 (good health and well-being). Our study enhances understanding of global development patterns and priorities and supports efforts to rescue the 2030 Agenda
Audit monitoring Human Immunodeficiency Virus (HIV) testing among girls and women at Heart of Hounslow Sexual Health Service
No full textThe gravitational wave background: observational prospects with the next generation experiments
No full textOver the past decade, gravitational-wave (GW) astronomy has transitioned from a theoretical frontier to a precision observational science, offering unprecedented insights into the most energetic phenomena in the universe. This thesis advances the study of the stochastic gravitational-wave background (SGWB)—the diffuse superposition of unresolved GW sources—across a broad frequency spectrum, spanning ground-based interferometers, space-based detectors, pulsar timing arrays (PTAs), and astrometric observatories.
We develop a unified formalism to model anisotropies and polarization in the SGWB, applicable to current and next-generation interferometric networks, such as the Einstein Telescope, Cosmic Explorer, LISA, and Taiji. This framework enables, for the first time, a systematic evaluation of the limits imposed by instrumental noise and cosmic variance on the measurement of anisotropies and polarization of the GW background.
The thesis further extends the formalism to non-interferometric probes of low-frequency GWs. We introduce a general framework to describe observables for nanohertz GWs surveys. We propose a novel astrometric technique to probe time-dependent shape distortions in resolved astrophysical sources induced by gravitational waves and short-distance astrometry. The predicted two-point correlation functions for this astrometric shear effect mirror the Hellings–Downs signature from PTAs and open a novel observational window using upcoming surveys such as Gaia and the Vera C. Rubin Observatory’s LSST.
By integrating comprehensive modeling of GW observables, response functions and innovative detection strategies, this work provides a general framework on the mathematical and statistical tools that will be crucial in the next era of GW science. Its findings support the science case and analysis pipelines of current and future GW observatories, with implications for astrophysics, cosmology, and the physics of the early universe.Open Acces
Tempo: compiled dynamic deep learning with symbolic dependence graphs
No full textDeep learning (DL) algorithms are often defined in terms of temporal relationships: a tensor at one timestep may depend on tensors from earlier or later timesteps. Such dynamic dependencies (and corresponding dynamic tensor shapes) are difficult to express and optimize: while eager DL systems support such dynamism, they cannot apply compiler-based optimizations; graph-based systems require static tensor shapes, which forces users to pad tensors or break-up programs into multiple static graphs.
We describe Tempo, a new DL system that combines the dynamism of eager execution with the whole-program optimizations of graph-based compilation. Tempo achieves this through a declarative programming model with recurrent tensors, which include explicit temporal dimensions. Temporal dimensions can be indexed using symbolic expressions to express dynamic dependencies on past and future tensors. Based on this, Tempo constructs a symbolic dependence graph, which concisely encodes dynamic dependencies between operators, and applies whole-program optimizations, such as algebraic simplifications, vectorization, tiling, and fusion. By tiling dynamic dependencies into static-size blocks, Tempo can also reuse existing static code-generators. It then uses a polyhedral model to find a feasible execution schedule, which includes memory management operations. We show that Tempo achieves a 7× speedup over JAX for Llama-3.2-3B decoding; for reinforcement learning algorithms, Tempo achieves a 54× speedup, with 16× lower peak memory usage
Designing and maintaining an efficient webassembly mechanisation
No full textWebAssembly is a rapidly evolving binary format originally designed for efficient and safe execution of low-level languages like C and Rust on the web supported by all major browsers. Since its release in 2017, it has expanded to diverse execution environments and has grown extensively in its features. The importance of the language and its design with a formal specification makes it a compelling subject for programming languages and formal verification research.
This thesis focuses on my contributions to studies around the WebAssembly semantics and verification. I contributed a major part in the WasmCert-Coq mechanisation of the WebAssembly in Coq, starting from its inception which drew definitions from Watt's Isabelle/HOL mechanisation for a pre-1.0 version of WebAssembly, until my update of the semantics to Wasm 2.0 and beyond, which also produces a fully conformant interpreter as its extracted artifact. My work on the WebAssembly mechanisation has discovered and helped fix several errors in the WebAssembly specification, fulfilling the promise of the language mechanisation for refining an important non-mechanised specification.
In addition, through my work on WebAssembly mechanisation, several methods have been explored and implemented to reduce the effort of maintaining mechanised models of WebAssembly. In particular, my work on the progressful interpreters proposes a new method using a dependently-typed design to merge the interpreter, its correctness, and the progress property into one mechanisation artifact.
Besides designing the mechanisation of WebAssembly’s semantics, I demonstrated the usefulness of language mechanisation to the formal verification by jointly developing the Iris-Wasm program logic using the Iris separation logic framework, which supports modular reasoning about WebAssembly programs, including host-WebAssembly interactions and module instantiation.
My efforts provide a foundation for efficient, scalable, and reliable verification of WebAssembly as it continues to evolve, ensuring its role as a secure and efficient platform for modern computation.Open Acces
Contribution of ethnicity and deprivation to paediatric critical care outcomes in the UK, 2008–21: a national retrospective cohort study
No full textBackground
Evidence from UK paediatric intensive care units (PICUs) demonstrates increased incidence of admission among children of Asian and Black ethnicity and children residing in more deprived areas. We aimed to investigate whether mortality in PICU is associated with ethnicity and child poverty.
Methods
This national cohort study included children aged 0–15 years who were admitted to PICUs in the UK between Jan 1, 2008, and Dec 31, 2021. Participating PICUs were affiliated with and reported outcome data to the Paediatric Intensive Care Audit Network (PICANet), from which we extracted data for this analysis. Key exposures were ethnicity, categorised as Asian, Black, Multiple, Other, and White, and area-level deprivation at the time of admission, quantified with the Children in Low-Income Families measure of an area's proportion of children living in families with income less than 60% of the median income and receiving out-of-work benefits or tax credits. The primary outcome was mortality during PICU stay. Mixed-effects logistic regression clustered by admitting PICU centre and individual patient was used to examine the association between key exposures and mortality during PICU stay for all admissions and then separately for planned and unplanned admissions.
Findings
This analysis included 245 099 admissions for 163 163 children during the study period, during which 102 990 (63·1%) had unplanned PICU admissions. 15 017 (9·2%) admissions were for children of Asian ethnicity, 7244 (4·4%) for children of Black ethnicity, 4514 (2·8%) for children of multiple ethnicities, 3831 (2·4%) for children of other ethnicities, and 100 241 (61·4%) for children of White ethnicity. Ethnicity data were missing for 32 316 (19·8%) of admissions. Observed PICU mortality across planned and unplanned admissions was 3·7% (9056 deaths per 245 099 admissions). Crude PICU mortality was highest among Asian children (1336 [5·1%; 95% CI 4·9–5·5] deaths per 26 022 admissions) and lowest among White children (4960 [3·2%; 3·1–3·3] deaths per 154 041 admissions), indicating a higher relative odds of PICU mortality among Asian children than White children for all admission types (odds ratio [OR] 1·52 [95% CI 1·42–1·62]). Odds of PICU mortality did not differ between children of Black and White ethnicity (OR 1·04; 95% CI 0·93–1·15). Children of multiple ethnicities (OR 1·23 [95% CI 1·08–1·39]) and other ethnicities (1·20 [1·05–1·38]) showed increased odds of PICU mortality relative to White children. PICU mortality ranged from 1025 (3·1%; 95% CI 2·9–3·3) deaths per 33 331 admissions for children in the least deprived quintile to 2432 (4·2%; 4·0–4·4) deaths per 58 110 admissions among children in the most deprived quintile, demonstrating an association between PICU mortality and area-level child poverty (OR 1·13 [95% CI 1·03–1·23]).
Interpretation
Variation in PICU mortality by ethnicity and area-level deprivation highlights the importance of further investigation into systemic drivers of disparities in PICU outcomes in the UK. Targeted strategies to mitigate drivers of inequalities are essential
Accelerating cDTI with deep learning-based tensor de-noising and breath hold reduction. a step towards improved efficiency and clinical feasibility
No full textBackground
Cardiac Diffusion Tensor Imaging (cDTI) non-invasively provides unique insights into cardiac microstructure. Current protocols require multiple breath-hold repetitions to achieve adequate signal-to-noise ratio, resulting in lengthy scan times. The aim of this study was to develop a cDTI de-noising method that would enable the reduction of repetitions while preserving image quality.
Methods
We present a novel de-noising framework for cDTI acceleration centred on three fundamental advances: (1) a paradigm shift from image-based to tensor-space de-noising that better preserves structural information, (2) an ensemble of Vision Transformer-based models specifically optimised for tensor processing through adversarial training, and (3) a sophisticated data augmentation strategy that maximises training data utilisation through dynamic repetition selection.
Results
Our approach reduces scan times by a factor of up to 4 while achieving a 20% reduction in cDTI maps errors over existing de-noising methods (Table 1) and preserving anatomical features such as infarct characterisation and transmural cardiomyocyte orientation patterns. Crucially, our proposed method succeeds in clinical cases where other algorithms previously failed.
Conclusions
This demonstrates substantial improvements in cDTI acquisition efficiency, achieving up to 4-fold scan time reduction (3-5 breath-holds) while maintaining diagnostic accuracy across diverse cardiac pathologies