21090 research outputs found
Sort by
Software Infrastructure for Isolation and Performance Monitoring in Virtualized Systems
No full textModern multiprocessor System-on-Chip (SoC) architectures host a rich tapestry of heterogeneous components, enabling multiple workloads with differing requirements to run simultaneously on the same hardware platform. However, managing and isolating these concurrently running applications presents significant challenges. Traditional virtualization techniques, even with static partitioning hypervisors, could struggle to ensure robust isolation due to contention in shared system resources such as caches and memory bandwidth. To address this issue, this thesis investigates memory bandwidth contention among cores and explores isolation strategies by implementing MemGuard in the Bao Hypervisor on ARMv8-based systems. This implementation is complemented by cache coloring and DRAM bank partitioning techniques. The results, evaluated using the San Diego Vision Benchmark Suite, quantify the effectiveness of these mechanisms in reducing interference and provide insights into program behavior under varying isolation parameters.
Beyond improving isolation, performance monitoring must extend beyond core-level observation to encompass system-wide interactions. To this end, this thesis develops a comprehensive software infrastructure for an Advanced Performance Monitoring Unit (APMU), designed for event-driven monitoring and dynamic runtime reconfiguration. By leveraging an LLVM-based toolchain to support custom instructions and integrating seamlessly with the hypervisor and guest OS layers, the APMU framework enables diverse applications while optimizing memory utilization and execution time.
Collectively, the results and infrastructure presented in this work contribute to more predictable, secure, and efficient computing systems, advancing the state of the art in virtualization, performance isolation, and heterogeneous system analysis
The Aesthetics of Resistance in Australian-run Immigration Detention Centres on Manus Island: The Case Study of Behrouz Boochani
No full textThe offshore detention regime for asylum seekers represents a contested model of border control,
punishing those seeking refuge and violating their fundamental rights. This research, however,
reveals a powerful counterpoint: the use of art and creativity as tools of resistance by detainees.
Through the story of Behrouz Boochani, an Iranian-Kurdish journalist and former Manus Island
detainee, this study illuminates the power of art and creativity in challenging this system. Despite
his lengthy confinement, Boochani produced a remarkable body of creative work that exposed
the harsh realities of detention, creating a counter-narrative that gained international acclaim.
This research introduces creativity as a tool for political activism, challenging the invisibility
inherent in offshore detention. The concept of ‘creative subjectivation’; is presented as an
analytical framework to understand how creative practices facilitate the transformation of
marginalized refugees into active political subjects. To explore this, the study investigates the
structural, operational, and experiential dimensions of offshore detention, from macro-level
border policies to micro-level dynamics within detention centers. Drawing on qualitative in-
depth interviews with Boochani, his creative collaborators, journalists, human rights advocates,
and former detainees, this study provides a multifaceted perspective on creative resistance within
highly restrictive spaces of border enforcement. The dissertation comprises seven chapters that
explore themes of border politics, the evolution of Australia’s offshore detention policies, the
lived experiences within detention centers, and the transformative potential of creative resistance
and includes the production of a documentary film that offers an immersive and sensorial
exploration of creative resistance and migrant activism within the offshore detention regime.
This project contributes to critical migration and border studies by illuminating the
transformative potential of creative resistance in contexts of extreme marginalization. It offers
new insights into refugee agency, migrant politics, border politics, and the role of art in
contesting anti-asylum policies and practices
Discovery and characterization of novel biofilm-associated proteins in Pseudoalteromonas tunicata
No full textPseudoalteromonas tunicata is a marine bacterium that is a useful model for studying mechanisms of biofilm development due to its ability to form, colonize, and inhibit growth of other microorganisms in marine and eukaryotic host-associated biofilms. However, the pathways responsible for P. tunicata biofilm formation are still incompletely understood, in part due to a lack of functional information for a large proportion of its proteome. In this thesis, I use comparative shotgun proteomics to explore P. tunicata biofilm development from the planktonic phase to three stages (early, middle, late) of biofilm development. Proteomic analysis identified 232 proteins that were up regulated during different stages of biofilm development, including proteins known to be important for P. tunicata biofilm development (e.g., autocidal enzyme AlpP, violacein proteins, and various pili proteins) as well as many hypothetical proteins of unknown function. I then characterized two novel, biofilm-associated hypothetical proteins, labeled EAR28894 and EAR30327.
Functional characterization of EAR28894 revealed that it is the major S-layer protein of P. tunicata. Bioinformatic methods predicted a beta-helical structure for EAR28894 similar to the Caulobacter S-layer protein, RsaA, despite sharing less than 20% sequence identity. Transmission electron microscopy revealed that purified EAR28894 protein assembled into paracrystalline sheets with a unique square lattice symmetry and a unit cell spacing of ~9.1 nm. An S-layer was found surrounding the outer membrane in wild-type cells and completely removed from cells in an EAR28894 deletion mutant. S-layer material also appeared to be “shed” from wild-type cells and was highly abundant in the extracellular matrix where it is associated with outer membrane vesicles and other matrix components. EAR28894 and its homologs form a new family of S-layer proteins that are widely distributed in Gammaproteobacteria including species of Pseudoalteromonas and Vibrio and found exclusively in marine metagenomes. This novel protein family was given the name Slr4.
Functional investigation of the uncharacterized protein, EAR30327, revealed its function as a novel biofilm adhesin. This protein, which I designated as BapP, was the top identified biofilm-associated protein by proteomic analysis. BapP showed partial homology to outer membrane adhesins containing repeats of bacterial cadherin-like and immunoglobulin (Ig) domains. A ΔbapP mutant strain was unable to form proper pellicle biofilms in liquid media. The Δ bapP mutant also had a significantly reduced ability to form biofilms in crystal violet assays, which was rescued by re-insertion of the bapP gene into the genome. As predicted by the identification of putative Ca2+-binding motifs in BapP, biofilm formation in the wild-type strain was demonstrated to be Ca2+-dependent, which was significantly reduced in the ΔbapP mutant. This study provides a unique proteomic dataset of biofilm development and identifies BapP as a Ca2+-dependent adhesin responsible for biofilm formation in P. tunicata. The occurrence of BapP-related homologs in other species suggests that this protein family represents a broadly conserved mechanism for biofilm adhesion in marine Gammaproteobacteria species.
This thesis research establishes a proteomics-based pipeline for biofilm protein discovery and new directions for biofilm research in P. tunicata and related bacteria, and offers insights into potential targets for biofilm management and control
Planting Imagination: Community Co-Design 'How-to-Guide'
No full textThis How-to Guide details how a group of architects, residents, community organizers, and public health researchers embarked on a journey to explore more radical ways of designing, working, and building together. Our aim was to go beyond the usual (and at times disempowering) community consultation methods commonly used in neighbourhood design, to see if we can develop an ethical co-design approach that starts and ends with the local community.
Through a combination of virtual reality (VR) centric visioning sessions and co-design workshops, we worked together to redesign Cecil’s community garden located in Toronto’s Chinatown West. This process helped us build community power to address ‘neighbourhood health’ as a collective resource in the context of COVID-19, and encouraged us to think beyond individualized health outcomes.
This guide is our way of sharing what we learned. We hope to encourage others to design and build in a more authentic and generative way by collaborating with local communities, and be open to bringing together diverse skills, practices, knowledge, and people. We’re excited to see how you might take, build, and make use of our insights in your own work and community. Let’s get started!本操作指南将详细介绍一群建筑师、居民、社区组织者和公共卫生研究人员如何踏上探索更激进的共同设计、工作和建设方式的旅程。我们的目标是超越社区设计中常用（有时让人感到无能为力）的社区咨询方法，看看我们能否开发一种有道德和以当地社区为起与终点的共同设计方法。
通过以虚拟现实 (VR) 为中心的愿景构想会议和共同设计活动，我们協力改造了位于多伦多西区唐人街思豪社区花园的设计。这过程帮助了我们建立社区力量以呼应在COVID-19背景下将“社区健康”作为集体资源看待，并鼓励我们进行超越个人健康结果的思考。
本指南是我们分享所学的方式。我们希望鼓励其他人通过与当地社区合作来达到更真实，更具创造性的方式进行设计和建造，并接纳将不同的技能、实践、知识和人才汇集在一起。我们很期待看到您如何在自己的工作和社区中收取，构建和利用我们的见解。让我们开始吧！This ‘how-to-guide’ was supported in part by funding through the University of Waterloo SSHRC Institutional Grant (SIG) (SSHRC Exchange 3). Planting Imagination was supported in part by funding from the Government of Canada’s New Frontiers in Research Fund (NFRF) through the tri-agency (CIHR, NSERC and SSHRC), Myseum of Toronto and Toronto Metropolitan University
Coupling Metabolic and Hydrodynamic Compartmental Models for Bioreactor Simulations
No full textLarge-scale bioreactors are widely employed across bioprocessing industries for the production of chemicals, pharmaceuticals, and biofuels. The increasing demand for specialized pharmaceuticals has motivated industries to optimize bioreactor operations. However, the complexity of multiphase interactions and the emergence of concentration gradients and intracellular heterogeneity in bioreactors pose significant challenges in accurately predicting and optimizing the performance of bioreactors. Simulations of numerical models have become invaluable for understanding these systems; however, the high computational cost of detailed models—particularly those involving multiphysics—limits their practicality. The computational cost of these models precludes them from being used for real-time applications or for extensive design optimizations.
To address this challenge, this thesis proposes computationally efficient methods to solve coupled metabolic and hydrodynamic compartmental models that describe key process dynamics.
The compartmental model (CM) approach is based on steady-state multiphase computational fluid dynamic simulations and is designed to accurately represent hydrodynamic properties such as turbulent dissipation energy, oxygen mass transfer, and flow topology. Conventional compartmentalization methods were found to introduce nonphysical ``short-cutting'' effects, leading to inaccuracies in mixing time predictions. To mitigate this, a refined compartmentalization approach was developed to better capture hydrodynamic mixing patterns. Then, a metabolic model was integrated with the compartmental model to explore the metabolic and hydrodynamic interactions.
In terms of the metabolic behavior, two key scenarios were considered: i) the intracellular concentrations were assumed to reach instantaneous equilibrium with their extracellular environments at all times and ii) the intracellular concentrations were not in equilibrium with their extracellular environments.
For the first case, where intracellular and extracellular equilibration was assumed, a binary search tree metabolic model (BSTMM) was developed from a dynamic flux balance analysis model and coupled with a CM describing the extracellular environment. This method significantly reduces computational complexity by substituting traditional linear optimization solvers with an online point-location approach. The coupled BSTMM-CM successfully captured diauxic growth dynamics and demonstrated substantial computational efficiency, enabling long-term bioprocess simulations on standard desktop hardware. However, kinetic parameters for metabolic models calibrated in small-scale bioreactors could not be directly applied to large-scale systems without recalibration. This finding suggested that intracellular heterogeneity may play a crucial role in metabolic regulation and must therefore be explicitly accounted for.
To address this, the second case considered a finite-rate adaptation mechanism governing equilibration between the extracellular and intracellular environments. A method of moments approach, assuming a truncated normal distribution, was implemented to reconstruct the number density function efficiently. This approach demonstrated that intracellular heterogeneity is most pronounced when the characteristic timescales of microbial adaptation and extracellular advection are comparable. The application of this approach to E. coli fermentation data reported in the literature resulted in improved fitting as compared to a model that ignores intracellular heterogeneity.
Furthermore, the impact of population heterogeneity on metabolic regulation was evaluated, revealing distinct variations in growth rate and substrate uptake across different regions of the bioreactor.
To further improve computational efficiency of the coupled PBM-CM, an adaptive population compartmental scheme is proposed, which dynamically adjusts the compartmentalization over the course of a simulation to balance accuracy and computational cost. This approach was found to be particularly effective for large-scale bioreactor simulations, especially when advection rates exceed microbial adaptation rates, leading to substantial reductions in simulation time with minimal loss of predictive accuracy.
This research significantly advances the modeling of large-scale bioreactors by integrating hydrodynamic and metabolic models into a computationally efficient framework. The developed methods provide more in-depth insights into the influence of concentration gradients and intracellular heterogeneity on microbial behavior, ultimately improving the predictive accuracy and scalability of bioprocess simulations
Reconstructing Late Holocene Environmental Change in the Pevensey Levels: Stratigraphic and Paleoenvironmental Insights from Horse Eye and North Eye, UK
No full textThis thesis investigates the stratigraphic framework and paleoenvironmental evolution of the Pevensey Levels, focusing on the depositional histories of the sediment covering the bedrock-cored islands Horse Eye and North Eye in East Sussex, United Kingdom (UK). Low-lying coastal systems such as the Pevensey Levels are sensitive to climatic variability, geomorphological changes, and human interaction, yet detailed stratigraphic and environmental reconstructions are limited for these landscapes. By combining field observations, sedimentological analyses, and laboratory analyses, this research enhances the understanding of climatic events, geomorphological factors, and anthropogenic influences that have shaped this low-lying coastal-lagoonal landscape over the Holocene.
A multiproxy laboratory approach- including laboratory methods that include laser diffraction grain size analysis, portable X-ray fluorescence (pXRF), loss on ignition (LOI), X-ray diffraction (XRD), and AMS radiocarbon dating- was applied to create a stratigraphic framework for reconstructing the stratigraphy and environmental history of the sediment on these bedrock-cored islands. The stratigraphic analyses reveal a broadly consistent depositional sequence across both islands, transitioning from lower marine silty clays at the base to organic-rich peat layers and finally to upper terrestrial clayey silts. Despite this similarity, there are differences that emerge: North Eye’s stratigraphy includes sand-dominated units, attributed to its thinner sediment cover, bedrock exposure, and localized sediment contributions during episodic higher-energy depositional events. In contrast, Horse Eye’s thicker sediment cover and more continuous peat layers indicate prolonged lower-energy deposition and water-logged conditions. Additionally, the lithologic and stratigraphic analyses conducted in this thesis offer a higher level of detail compared to earlier studies, providing localized depositional variability. These findings distinguish this research from earlier work in the area, which provided generalized stratigraphic data for North Eye and focused on broader regional depositional sequences within the Pevensey Levels.
The integration of results from the different techniques identified a temporal alignment between localized responses in the sediment with key climatic events, such as the globally recognized 4.2 ka event- a period characterized by dry climates in some regions but has remained poorly understood in Western Europe. At the Pevensey Levels, the stratigraphy during this period reveals an increase in sand content, indicative of heightened depositional energy and fluctuating hydrological conditions, providing new insights into how this global climatic phase may have influenced low-lying coastal landscapes in the UK. The findings also align temporally with documentation of human modifications during the late Holocene, including drainage and land reclamation. For example, lenticular laminations and an increase in sand content in the upper stratigraphy align with documented medieval drainage efforts in this region. This thesis situates these modifications within a regional context, noting how anthropogenic activities may have influenced sedimentary processes in dynamic coastal environments.
This thesis provides a new framework for understanding the evolution of the Pevensey Levels within a broader regional context by drawing comparisons with adjacent systems such as the Romney Marsh and Somerset Levels. While the Pevensey Levels exhibit broadly similar depositional patterns to adjacent systems, including marine-to-terrestrial transitions, there are localized differences in sediment depositional processes due to geomorphological controls, such as the differences in size and shape of the bedrock islands, as well as their location with respect to their proximity to sediment sources. The Pevensey Levels’ stratigraphy is more influenced by bedrock-controlled sedimentation near the bedrock-cored islands, landforms that are not present in the Romney Marsh. These regional comparisons reveal variations in depositional energy, peat development, and anthropogenic modifications, offering new insights into the factors shaping coastal-lagoonal systems during the Holocene
Advances in the Analysis of Irregular Longitudinal Data Using Inverse Intensity Weighting
No full textThe analysis of irregular longitudinal data can be complicated by the fact that the timing at which individuals are observed in the data are related to the longitudinal outcome. For example, this can occur when patients are more likely to visit a clinician when their symptoms are worse. In such settings, the observation process is referred to as informative, and any analysis that ignores the observation process can be biased. Inverse intensity weighting (IIW) is a method that has been developed to handle specific cases of informative observation processes. IIW weights observations by the inverse probability of being observed at any given time, and creates a pseudopopulation where the observation process is subsequently ignorable. IIW can also be easily combined with inverse probability of treatment weighting (IPTW) to handle non-ignorable treatment assignment processes. While IIW is relatively intuitive and easy to implement compared to other existing methods, there are few peer-reviewed papers examining IIW and its underlying assumptions.
In this thesis, we begin by evaluating a flexible weighting method which combines IIW and IPTW through multiplication to handle informative observation processes and non-randomized treatment assignment processes. We show that the FIPTIW weighting method is sensitive to violations of the noninformative censoring assumption and show that a previously proposed extension fails under such violations. We also show that variables confounding the observation and outcome processes should always be included in the observation intensity model. Finally, we show scenarios where weight trimming should and should not be used, and highlight sensitivities of the FIPTIW method to extreme weights. We also include an application of the methodology to a real data set to examine the impacts of household water sources on malaria diagnoses of children in Uganda.
Next, we investigate the impact of missing data on the estimation of IIW weights, and evaluate the performance of existing missing data methods through empirical simulation. We show that there is no "one-size-fits-all" approach to handling missing data in the IIW model, and show that the results are highly dependent on the type of covariates that are missing in the observation times model. We then apply the missing data methods to a real data set to estimate the association between sex assigned at birth and malaria diagnoses in children living in Uganda.
Finally, we provide an in-depth evaluation on the assumptions made on IIW across various peer-reviewed papers published in the literature. For each set of assumptions, we construct directed acyclic graphs (DAGs) to visualize the assumptions made on the observation and censoring processes which we use to highlight inconsistencies and potential ambiguity among the assumptions presented in existing works involving IIW. We also discuss when causal estimates of the marginal outcome model can be obtained, and propose a general set of assumptions for IIW
Design and Development of a Real-time Monitoring Microfluidic Platform for Multiplexed Biomarker Detection
No full textContinuous, multiplexed, real-time measurements of biomarkers can unravel useful information about different health-related problems in patients. In the case of diabetes and obesity, glucose can be rapidly, repeatedly, and now continuously measured. The time course of changes in blood glucose after a glucose intake is well-established, and there has also been progress in defining glucose responses to different foods and how individual glucose responses predict and change in prediabetes and type 2 Diabetes (T2D). However, compared to blood glucose, very little is known about dynamic changes in blood insulin responses, especially highly time-resolved postprandial changes in insulin and related hormones/peptides such as C-peptide and glucagon.
In addition to blood glucose, there is a need to measure highly time-resolved insulin, glucagon, and C-peptide responses using continuous monitoring. This is an important knowledge gap because current methods to assess these hormones/peptides only capture a few time points after ingesting meal or glucose intake, which is a serious limitation in both preclinical and human assessments. In this work, a multi-module microfluidic-based platform is developed and utilized to go beyond solely glucose measurements. This system, called quantum dot integrated real-time ELISA (QIRT-ELISA), measures glucose, insulin, glucagon, and C-peptide levels continuously and simultaneously.
First, a sensitive bead-based quantum dot immunoassay (BQI) has been validated for insulin and glucagon detection. The integration of the BQI facilitated multiplexed and continuous monitoring. Moreover, the QDot technology used in BQI assists with the enhancement in the sensitivity of the immunoassays for both insulin and glucagon. Next, the QIRT-ELISA system was validated for in vitro continuous measurements of insulin and glucagon in whole blood samples without a need for pre-processing the sample. Finally, the results from discrete glucose tolerance tests (GTT) conducted by the developed microfluidic platform from in vivo rat models showed successful cross-validation of the device with the gold-standard ELISA.
In the next step, the QIRT-ELISA was expanded for more complex system measurements. The expanded QIRT-ELISA can monitor four biomarkers continuously in a multiplexed setting. To bridge the gap in diabetes studies, insulin, glucagon, C-peptide, and glucose were selected as the targets of the study. The BQI was employed for insulin, glucagon, and C-peptide, and a new aptamer bead-based assay was developed for glucose measuring. All assays were tested for their specificity against their target since the number of biomarkers was increased in this step of the work. Next, QIRT-ELISA was tested for its ability for continuous and multiplexed in vivo monitoring of insulin, glucagon, C-peptide, and glucose on conscious rat models in a GTT experiment.
With the continuous measurements of all four biomarkers, the QIRT-ELISA provides more data compared to conventional ELISA, which enables us to explore new interplays between the biomarkers under various scenarios.
The results obtained by this system will assist with shedding light on fundamental knowledge gaps of how factors beyond blood glucose are involved in the progression of pre-diabetes and T2D and add vital information to continuous glucose monitoring for precision nutrition, which finally leads to better diabetes management.
The clinical field requires a reliable approach that ensures continuous, cost-effective, and high-throughput separation of blood plasma, achieving sufficient purity for the accurate detection of biomarkers. Current methods, such as gold-standard centrifugation and microfluidic technologies, do not adequately fulfill these requirements. In the current work, we updated a passive hydrodynamic device to maintain high yield while achieving admissible purity, and high-quality plasma samples.
Through both a computational model and experimental trials, we optimized the employed side channels’ lengths, which contributed to improving the plasma extraction rate significantly. These optimized side channels utilize the established cell-free layer areas within the contraction expansion regions to facilitate the reliable and effective extraction of plasma. Named Hydrodynamic Continuous, High-Throughput Plasma Separator (HCHPS) microfluidic device, the optimized device works at high throughput and achieves a purity of 47% with whole and 62% with 1:1 diluted blood, while increasing the yield by three times compared to the previous studies.
Finally, the optimized device was used to extract lactate-containing plasma from whole blood samples and was compared to plasma separated by centrifugation by conducting a bead-based fluorescence biosensing and an electrochemical aptamer-based biosensing. With comparable results from this experiment, we showed that this device can potentially be integrated with other microfluidic platforms for more sensitive downstream analysis of whole blood samples
Chemo-rheological Characterization of Asphalt Binders Using Different Aging Processes
No full textThe performance and longevity of asphalt pavements depend heavily on the properties of asphalt binders, which are affected by aging, binder modifications, and the incorporation of reclaimed asphalt pavement (RAP) materials. However, significant gaps exist in understanding the long-term chemical and rheological changes induced by aging processes (particularly with respect to differences between thermo-oxidative aging and UV exposure), and in the use/standardization of chemical analytical techniques such as Fourier Transform Infrared (FTIR) and Nuclear Magnetic Resonance (NMR) spectroscopy for binder characterization. Furthermore, the behaviour in RAP-virgin binder blends, along with the influence of bio-based rejuvenators and anti-aging additives under different aging conditions, remains underexplored. Addressing these gaps are crucial to developing more durable, sustainable pavements. This thesis bridges these research gaps through comprehensive investigation of chemo-rheological binder characterization, combining experimental testing with advanced analytical tools and varying aging methods. The findings offer essential insights into binder aging, rejuvenation strategies, and modification techniques, with significant implications for pavement durability and environmental sustainability.
The first chapter presents an evaluation of Attenuated Total Reflection-Fourier Transform Infrared (ATR-FTIR) spectroscopy combined with functional group and multivariate analysis techniques to characterize asphalt binders. The research identifies challenges in repeatability across binder sources and aging states demonstrating the importance of standardized protocols for improving reliability. Repeatability as described by AASHTO standards is listed in the precision and bias statement as single operator precision. This is the allowable difference in two test results measured under the repeatability conditions (same asphalt binder, measured by the same operator, on the same piece of equipment in the same lab). Principal Component Analysis (PCA) and k-means clustering successfully classified binder types and aging states, with large quantity (LQ) sample preparation yielding more consistent results than small quantity (SQ) preparation. These findings underscore the need for uniform procedures in binder analysis, addressing inconsistencies prevalent in the current literature.
The second part of the thesis investigates the impact of Styrene-Butadiene-Styrene (SBS) polymer modification on binder performance and oxidative resistance. Using Nuclear Magnetic Resonance (NMR) and ATR-FTIR spectroscopy, along with PCA and Partial Least Squares Regression (PLSR), the research highlights the ability of SBS to enhance high-temperature performance and slow thermo-oxidative aging. This work not only confirms previous findings on SBS but also provides new insights into the molecular interactions contributing to aging resistance. The study fills a gap in understanding how SBS-modified binders behave under various aging scenarios, offering a deeper perspective on polymer-modified asphalt technologies.
The thesis also addresses a critical gap related to UV-induced aging, which has been underexplored in comparison to thermo-oxidative aging. A novel UV aging chamber was developed to simulate real-world environmental conditions, incorporating UV exposure, water spray cycles, and controlled heating at 70°C. Comparative analysis revealed that different additives exhibit varying effectiveness under UV and thermo-oxidative conditions. Zinc diethyldithiocarbamate (ZDC) showed strong resistance to thermo-oxidative aging but limited efficacy under UV aging, while ascorbic acid (Vit. C) accelerated aging under UV exposure, contrary to expectations. These findings emphasize the challenges involved in designing effective anti-aging strategies for asphalt binders, demonstrating the value of combining conventional rheological tests with spectroscopic techniques and further highlighting the need for more targeted approaches to additive selection and development.
This thesis advances the understanding of asphalt binder behaviour and aging processes by integrating chemical, rheological, and multivariate analysis techniques. It offers critical contributions to the standardization of binder characterization protocols, the optimization of polymer-modified asphalt technologies, and the development of more effective anti-aging strategies. The research also demonstrates the potential of machine learning and artificial intelligence (AI) in predicting binder performance from spectroscopic data using multivariate analysis, paving the way for future innovations in asphalt binder characterization.
In conclusion, the work in this thesis addresses significant gaps in the literature, providing new insights into aging mechanisms, additive/rejuvenation strategies, and RAP binder interactions. By combining chemical analysis, rheological testing, and multivariate techniques, this research contributes both to academic knowledge and practical pavement engineering, promoting the development of more sustainable, long-lasting asphalt pavements
Code Generation and Testing in the Era of AI-Native Software Engineering
No full textLarge Language Models (LLMs) like GPT-4 and Llama 3 are transforming software development by automating code generation and test case creation. This thesis investigates two pivotal aspects of LLM-assisted development: the integration of Test-Driven Development (TDD) principles into code generation workflows and the limitations of LLM-based test-generation tools in detecting bugs.
LLMs have demonstrated significant capabilities in generating code snippets directly from problem statements. This increasingly automated process mirrors traditional human-led software development, where code is often written in response to a requirement. Historically, Test-Driven Development (TDD) has proven its merit, requiring developers to write tests before the functional code, ensuring alignment with the initial problem statements. Applying TDD principles to LLM-based code generation offers one distinct benefit: it enables developers to verify the correctness of generated code against predefined tests. This paper investigates if and how TDD can be incorporated into AI-assisted code-generation processes. We experimentally evaluate our hypothesis that providing LLMs like GPT-4 and Llama 3 with tests in addition to the problem statements enhances code generation outcomes. We experimented with established function-level code generation benchmarks such as MBPP and HumanEval. Our results consistently demonstrate that including test cases leads to higher success in solving programming challenges. We assert that TDD is a promising paradigm for helping ensure that the code generated by LLMs effectively captures the requirements.
As we progress toward AI-native software engineering, a logical follow-up question arises: Why not allow LLMs to generate these tests as well?
An increasing amount of research and commercial tools now focus on automated test case generation using LLMs.
However, a concerning trend is that these tools often generate tests by inferring requirements from code, which is counterintuitive to the principles of TDD and raises questions about their behaviour when the flawed assumption of the code under test being correct is violated.
Thus we set out to critically examine whether recent LLM-based test generation tools, such as Codium CoverAgent and CoverUp, can effectively find bugs or unintentionally validate faulty code.
Considering bugs are only exposed by failing test cases, we explore the question: can these tools truly achieve the intended objectives of software testing when their test oracles are designed to pass?
Using real human-written buggy code as input, we evaluate these tools, showing how LLM-generated tests can fail to detect bugs and, more alarmingly, how their design can worsen the situation by validating bugs in the generated test suite and rejecting bug-revealing tests.
These findings raise important questions about the validity of the design behind LLM-based test generation tools and their impact on software quality and test suite reliability.
Together, these studies provide critical insights into the promise and pitfalls of integrating LLMs into software development processes, offering guidelines for improving their reliability and impact on software quality