Search

Cardiac diseases are among the most prevalent in the world. The scope and impact of these diseases on society are far-reaching. Therefore, developing tools to aid in identifying and understanding the underlying mechanisms of these diseases is an ongoing and vital field of research in which computational modelling has emerged as an essential tool. Using flexible, open-source software, and modern numerical methods can aid in this endeavour. In this thesis, I have developed a computational platform, for the simulation of bio-physically detailed cardiac tissue models with general cell models. This platform consists of significant additions to the open-source multi-physics finite element library Object Oriented Multi-physics Library (Oomph-lib) and allows for the novel application of Oomph-lib to cardiac modelling. These developments couple single-cell models to tissue and 2D or 3D anatomical models of the heart. The platform can simulate electrical excitation waves and mechanical contraction of cardiac tissue. In addition to this platform, I have also developed methods of applying adaptive operator splitting methods for the stable and efficient numerical solution of cardiac models which I show outperform other more commonly used methods. In conclusion, an efficient, scalable, open-source computational platform applicable to cardiac modelling has been developed and further areas for development have been investigated and noted that form a basis for further model development and validation.

Software development is intrinsically a human activity and the role of the development team has been established as among the most decisive of all project success factors. Prior research has proven empirically that team size and stability are linked to stakeholder satisfaction, team productivity and fault-proneness. Team size is usually considered a measure of the number of developers that modify the source code of a project while team stability is typically a function of the cumulative time that each team member has worked with their fellow team members. There is, however, limited research investigating the impact of these factors on software maintainability - a crucial aspect given that up to 80% of development budgets are consumed in the maintenance phase of the lifecycle. This research sheds light on how these aspects of team composition influence the structural attributes of the developed software that, in turn, drive the maintenance costs of software. This thesis asserts that new and broader insights can be gained by measuring these internal attributes of the software rather than the more traditional approach of measuring its external attributes. This can also enable practitioners to measure and monitor key indicators throughout the development lifecycle taking remedial action where appropriate. Within this research the GoogleCode open-source forge is mined and a sample of 1,480 Java projects are selected for further study. Using the Chidamber and Kemerer design metrics suite, the impact of development team size and stability on the internal structural attributes of software is isolated and quantified. Drawing on prior research correlating these internal attributes with external attributes, the impact on maintainability is deduced. This research finds that those structural attributes that have been established to correlate to fault-proneness - coupling, cohesion and modularity - show degradation as team sizes increase or team stability decreases. That degradation in the internal attributes of the software is associated with a deterioration in the sub-attributes of maintainability; changeability, understandability, testability and stability.

Air pollution is associated with various adverse health impacts and is identified as one of the leading risk factors for global disease burden. Further, air pollution is one of the pathways through which climate change could negatively impact health. Field studies have shown that air pollution has high spatiotemporal variability and pollutant concentrations vary substantially within neighborhoods. Characterizing air pollution at a fine-grained level is essential for accurately estimating human exposure, assessing its impact to human health, and further aiding localized air pollution policy. Air quality models are developed to estimate air pollution at locations and time periods without monitors, and these estimates are commonly used for exposure and health effects studies. Traditional land use regression [LUR] models are one of the cost-effective empirical air quality models. LUR typically relies on fixed-site measurements, GIS-derived variables with limited spatial resolution, and captures linear relationships. In recent years, innovative open-source imagery datasets and their associated features (e.g., street view imagery, remote sensing imagery) have emerged and show potential to augment or replace traditional LUR predictors. Such imagery data sources embody abundant information of natural and built environment features. Advanced computer vision techniques enable feature extraction and quantification through these extensive imagery datasets. The overarching objective of this dissertation is to investigate the feasibility of leveraging open-source imagery datasets (i.e., Google Street View [GSV] imagery, Landsat imagery, etc.) and advanced machine learning algorithms to develop image-based empirical air quality models at both local and national scale. The first study of this work established a pipeline of feature extraction through street view imagery sematic segmentation. The resulting street view features were used to predict street-level particulate air pollution for a single city. The results showed that solely using GSV-derived features can achieve comparable model fits as using traditional GIS-derived variables. Feature engineering improved model stability and interpretability through reducing spurious variables from potential misclassifications from computer vision algorithms. The second study further developed GSV-based models at national scale across multiple years. Random forest models were developed to capture the nonlinear relationship between air pollution and its impacting factors. The results showed that with sufficient street view images, GSV imagery alone may explain the variation of long-term national NO2 concentrations. Adding satellite-derived aerosol estimates (i.e., OMI column density) can significantly boost model performance when GSV images are insufficient, but the addition narrows when more GSV images are available. Our systematic assessment of the impact of image availability on model performance suggested that a parsimonious image sampling strategy (i.e., one GSV image per 100m grid) may be sufficient and most cost-effective for model development and application. Our third study explored the feasibility of combining street view and remote sensing derived features for national NO2 and PM2.5 modeling and projection at high spatial resolution. We found that GSV-based models captured both the highest and lowest pollutant concentrations while remote sensing features tended to smooth the air pollution variations. The results suggested that GSV features may have the capability to better capture fine-scale air pollution variability. The resulting air pollution prediction product may serve a variety of applications, including providing new insights into environmental justice and epidemiological studies due to its high spatial resolution (i.e., street level). Collectively, the result of this dissertation suggests that GSV imagery, processed with computer vision techniques, is a promising data source to develop empirical air quality models with high spatial resolution and consistent predictor variables processing protocol. Image-based features assisted with advanced ML approaches have the potential to greatly improve air quality modeling estimates, and successfully show comparable and even superior model performance than other modeling studies. Moreover, the ever-growing public imagery data sources are particularly promising for remote or less developed areas where traditional curated geodatabases are sparse or nonexistent.

In the rapidly expanding semiconductor industry, there is an increasing demand for skilled chip developers. Yet, the steep learning curve associated with Hardware Description Languages (HDLs) often acts as a significant barrier for students hoping to pursue a career in digital design. Drawing upon my experience as a HDL educator, which includes teaching Verilog to UCSB's IEEE student chapter and serving as a Teaching Assistant for UCSB's Verilog courses, I have meticulously developed and refined a comprehensive set of methods and resources for Verilog education. My objective encompassed two key facets: equipping students with quality industry-preparation and kindling passion for exploring hardware design. Through a strategic blend of approaches consisting of the integration of accessible open-source tools, the enforcement of popular coding style guides, the implementation of autograders for personalized feedback, and the incorporation of open-source IP blocks into lessons, students can attain proficiency in designing RTL (Register Transfer Level) for rigorously verified hardware systems. These strategies help reduce Verilog's steep learning curve while also expediting the introduction of more advanced topics in digital design and computer architecture. The methods and resources detailed in this thesis will prepare students for the expectations of the semiconductor industry, enhance their coding skills, and promote an accessible and engaging learning environment, ultimately meeting the growing demand for chip developers.

An Uncrewed Aerial Vehicle (UAV) is an airborne vehicle that has no people onboard and thus is either controlled remotely via radio signals or by autonomous capability. This thesis highlights the feasibility of using a bio-inspired miniature lighter than air UAV for indoor applications. While multicopters are the most used type of UAV, the smaller multicopter UAVs used in indoor applications have short flight times and are fragile making them vulnerable to collisions. For tasks such as gas source localisation where the agent would be deployed to detect a gas plume, the amount of air disturbance they create is a disadvantage. Miniature blimps are another type of UAV that are more suited to indoor applications due to their significantly higher collision tolerance. This thesis focuses on the development of a bio-inspired miniature blimp, called FishBlimp. A blimp generally creates significantly less disturbance to the airflow as it doesn’t have to support its own weight. This also usually enables much longer flight times. Using fins instead of propellers for propulsion further reduces the air disturbance as the air velocity is lower. FishBlimp has four fins attached in different orientations along the perimeter of a helium filled spherical envelope to enable it to move along the cardinal axes and yaw. Support for this new vehicle-type was added to the open-source flight control firmware called ArduPilot. Manual control and autonomous functions were developed for this platform to enable position hold and velocity control mode, implemented using a cascaded PID controller. Flight tests revealed that FishBlimp displayed position control with maximum overshoot of about 0.28m and has a maximum flight speed of 0.3m/s. FishBlimp was then applied to source localisation, firstly as a single agent seeking to identify a plume source using a modified Cast & Surge algorithm. FishBlimp was also developed in simulation to perform source localisation with multiple blimps, using a Particle Swarm Optimisation (PSO) algorithm. This enabled them to work cooperatively in order to reduce the time taken for them to find the source. This shows the potential of a platform like FishBlimp to carry out successful indoor source localisation missions.

The slowdown in Moore's law, coupled with the insatiable demand for compute from today's emerging applications, has created the need for hardware systems far beyond current capabilities. Electronic design automation (EDA) tools are now challenged to build chips that not only compensate for slow down in scaling (design-equivalent scaling) but also provide high performance for both ML and non-ML applications, which use a variety of new architectural techniques and operate under stringent performance constraints. However, conventional EDA tools and EDA research are plagued by several challenges. First, they involve computationally expensive analysis and optimizations which increase design time-to-market. Second, they are heuristic-based and often tradeoff speed for accuracy which results in suboptimal solutions. Third, the closed culture within the community, expensive EDA tools, and complexity of EDA tools that demands expert users have created a high barrier to entry slowing down research. This thesis explores machine learning (ML) and open-source philosophies to address these challenges in EDA tools for physical design. Every stage of physical design involves the use of optimization algorithms, which invoke computationally expensive analysis, to minimize power and area and maximize the performance of the circuit. Three critical analyses that are performed several times during physical design are power delivery network (PDN) analysis, thermal analysis, and timing analysis. Optimization algorithms require these analyses to be fast and accurate to generate solutions that can drive design-equivalent scaling. While these analyses and optimization problems have been explored since the early 1980s, the advent of ML brings in novel solutions and an open culture that creates a new wave in EDA. The first part of this thesis shows how ML transcends conventional EDA tool challenges for power-related problems by (i) enabling vast speedups for PDN and thermal analysis with fast inference with high accuracy,(ii) providing optimized correct-by-construction PDN solutions, and (iii) aiding in PDN benchmark and dataset generation. To speed up conventional PDN and thermal analysis, this thesis maps them into image-to-image and sequence-to-sequence translation tasks, which allows leveraging a class of ML models with an encoder-decoder-based generative (EDGe) architecture. Once trained, these ML models are transferable across designs synthesized within the same technology and packing solution. The networks perform millisecond analyses with negligibly small errors against commercial tools that require several hours. Further, these EDGe networks are leveraged to develop MAVIREC, which can rapidly sample a large set of input vectors and provide recommendations for voltage drop analysis. While analysis diagnoses problems, it must be coupled with optimization techniques to solve these problems. The thesis proposes OpeNPDN, an ML-based methodology for PDN optimization that employs a set of predefined templates that serve as potential building blocks for the PDN. The optimization algorithm involves building a PDN that meets voltage drop and electromigration constraints while minimizing the wiring resources used. A convolutional neural network (CNN) is trained to select an appropriate PDN template for each region on the chip based on power, bump, macro, and congestion distributions. The CNN-based optimization rapidly frees thousands of routing tracks in congestion-critical regions, while staying within the voltage drop and electromigration (EM) current density limits. Although ML has found success in the problems described above, a major challenge has been the unavailability of benchmarks for evaluating these solutions. The thesis develops BeGAN, an ML-based methodology for synthesizing synthetic PDN benchmarks that obfuscate intellectual property information. The approach applies generative adversarial networks (GANs) and transfer learning techniques to create realistic PDN benchmarks from a small set of real circuit data. BeGAN generates thousands of PDN benchmarks released in the public domain in four open-source technologies. The second part of the thesis develops ML-based solutions for timing analysis and optimization. A critical part of timing closure is accurate crosstalk-aware timing analysis. However, existing crosstalk-aware static timing analysis (STA) techniques are limited due to their imprecise use of heuristics. The thesis proposes XT-PRAGGMA, a tool that uses GPU-accelerated dynamic gate-level simulations and ML to eliminate inaccuracies in crosstalk-aware STA tools and accurately predict crosstalk-induced delays. The proposed technique is fast and reduces falsely-reported total negative slack by 70% compared to traditional STA-based techniques. In addition to accurate analysis techniques, timing optimization algorithms are critical to successful timing closure. Logic gate sizing is one such NP-hard-constrained optimization problem that aims to minimize the circuit implementation cost (typically, area or power), subject to timing constraints. This thesis presents RL-LR-Sizer, a tool that applies Lagrangian relaxation techniques to a novel reinforcement learning (RL) framework to perform autonomous gate sizing. The RL framework trains a relational graph convolutional network agent to perform timing optimization. RL-LR-Sizer outperforms conventional algorithms and moves the Pareto optimal front of the area-delay tradeoff curve to the left on designs in a 45nm technology. The last part of the thesis aims to address challenges related to high barriers to entry and the closed culture in the EDA community, which hinder reproducible research and the adoption of ML-based techniques. It contributes to the open ecosystem by developing novel open-source EDA tools. The first is TherMOS, an open-source thermal model to estimate self heating effects in advanced transistors. Modern transistors such as FinFETs and gate-all-around FETs suffer from excessive heat confinement due to their small size and three-dimensional geometries. This results in device self-heating, which can reduce speed, increase leakage, and accelerate aging. The thesis uses TherMOS to characterize the temperature for both the 7nm FinFET and 5nm GAAFET sub-structures and analyzes the impact of temperature on circuit performance and reliability. The second is RTA-simulator, an open-source pattern-dependent transient rapid thermal annealing (RTA) simulator for analyzing RTA-induced variations due to differences in die layout patterns. Unlike prior art, RTA-simulator considers the temperature-dependent thermal conductivity of silicon. The thesis leverages RTA-simulator to analyze RTA effects on a 16KB SRAM and suggest strategies for RTA-induced variation mitigation for a 7nm FinFET SRAM design. The third is PDNSim, an open-source PDN analysis tool integrated with OpenROAD, for estimating voltage drop and EM current densities. With scaling, power densities and wire parasitics are on an increasing trend making PDN analysis an integral part of physical design. For the OpenROAD project, PDNSim serves two purposes. First, it performs static voltage drop estimation and current density estimation at various stages of the physical design flow after placement. Second, it checks the connectivity of the power grid to ensure that every standard cell receives power (voltage) from the pin (bump) and that all power stripes are connected. The thesis demonstrates the results of PDNSim on several designs in both open-source and closed-source technologies, which are being run as a part of the OpenROAD project regression tests.

While most states in America have passed laws permitting harsher punishments for those convicted of hate (or bias) crimes, there has been no research to date on the adjudication of these defendants, including how legal and extralegal attributes of bias crime shape prosecutorial and judicial decision-making. This gap in research is likely due in part to the limitations of official data on bias crimes. Fortunately, new data on legal outcomes for bias homicide offenders who target victims because of their race, ethnicity, nationality, religion, sexual orientation, gender identity or homed status have become available from the open-source database known as the Bias Homicide Database (BHDB). Drawing from the BHDB and theoretical perspectives on court-actor decision-making, the current study quantitatively investigates the relationships between victim attributes and the severity of case dispositions in bias-motivated homicide cases. Findings suggest that victim attributes are significantly related to justice outcomes. This study sheds light on how bias crime defendants are adjudicated in American courts with implications for broader debates regarding the social value and legal utility of enhanced punishments for those convicted of crimes targeting persons because of their social group.

This thesis presents the software stack that is the Short to Medium Autonomous Delivery System (SMADS), a framework developed as a team effort that enables autonomous deliveries by robots on the University of Texas at Austin campus. The goal of this work is to achieve long-term robot autonomy with the particular use case of last mile delivery, all while integrating a novel, highly modular, open-source software architecture to support a heterogeneous fleet of robots and a mobile application to facilitate customer interaction. The contributions detailed in this thesis include development of such a framework, integration of individual development work with those of other team members, and results from deployment of the system over a five-day period.

Over the last few years, there has been an increased interest in the use of ultrasoundin a variety of research applications. Due to this increased interest, there is a needfor the development of customized transmission, reception, and processing algorithms,along with low-cost programmable ultrasound imaging devices that would be more desirablefor the variety of research applications. This thesis describes a programmable 64-channel system-on-chip (SoC)-based test bench that will be utilized for future researchand development on autonomous wearable and implantable medical ultrasound imagingapplications. The current setup consists of a custom transmitter, a custom high-voltage(HV)multiplexer, an off-the-shelf receiver, and embedded software on a Systemon Chip. This prototype results in a modular design that can be easily updated withimproved hardware and software for different applications. This thesis describes thecurrent system hardware and software design along with the initial imaging results ontissue phantoms.

Dual-arm robots have several unique capabilities that make them very attractive for many applications, such as task flexibility, and the ability to perform “two-handed” object manipulation tasks that other systems cannot. Additionally, humanoid systems have the ability to access workspaces and use tools designed for humans. However, in order for these dual-arm systems to be utilized in particularly challenging environments – such as human-shared environments, outer space, or disaster zones – these systems will need to become much more capable than they are today. Currently, the control of these systems is made difficult due to their complex planning, dynamics, and control, and current dual-arm robotic systems are nowhere near resilient enough to handle the uncertainty and risk that are found in these use cases. A separate issue (one which is found in other areas of robotics as well) is that the control architectures are seldom shared, and are so specialized and hardcoded to their particular use cases that they cannot be easily reused even when shared. Extensive development effort, therefore, goes into redeveloping new systems for each particular application with limited ability for reuse.

Abstract: Low catalyst content remains a key requirement for the commercialization of polymer electrolyte fuel cells (PEFCs) into the energy, transportation, and material handling sectors. Understanding the fundamental phenomena reducing fuel cell performance at various stages of operation play a major role in PEFC optimization and reducing catalyst content. In this thesis, high performance PEFCs with low to moderate Pt loadings (27 - 112 μgPt/cm2) have been fabricated using inkjet printing. To better understand thin low Pt content electrodes the PEFCs are tested at various back pressures, relative humidity, and oxygen partial pressures. The characterized PEFCs are then simulated using OpenFCST, an open-source FEM based fuel cell simulation framework, to aid in the investigation of key performance limiting phenomena. The simulations are obtained using a macro-homogeneous, non-isothermal MEA model where a multi-step reaction kinetics describes the oxygen reduction reaction. The model is then validated against the experimental Pt loading, ionomer loading, and oxygen partial pressure study. The successfully validated model highlighted key performance limitations between low Pt content and conventional (400 μgPt/cm2) loading electrodes, while also highlighting the possible phenomena dictating ionomer and oxygen partial pressure performance. A mathematical model is also developed allowing the simulation of functionally graded electrodes. The model is then used to aid experimental design. To validate the new model two functionally graded ionomer electrodes are fabricated, characterized, and compared to the simulated results.

Abstract: The focus of this work is on improving understanding of mass transport limiting phenomena occurring within the micro-structure of Polymer Electrolyte Fuel Cell (PEFC) Catalyst Layers (CL). Micro-scale models, namely agglomerates, are employed within multi-scale CL framework, to examine many phenomena, such as the conductivity of protons in Nafion ionomer, non homogeneous catalyst distributions, oxygen dissolution in ionomer thin films, polydisperse agglomerate structures, and proton transport mechanisms within water filled carbon porous media. Catalyst distribution within CL micro structures, oxygen dissolution in Nafion described as a non-equilibrium process, and the mechanism by which protons are transported within water filled carbon pores are all found to be significant and potential explanations for PEMFC mass transport limited behaviour. The developed simulation framework presented in this work, as part of OpenFCST, the Open source Fuel Cell Simulations Toolbox, can be used in conjunction with experimental methods to improve understanding of said phenomena, improving understanding and design of PEMFCs.

Abstract: A comprehensive open-source two-phase non-isothermal MEA model including the microporous layer and accounting for multi-step reaction kinetics for the ORR and HOR is developed for the first time. The model is integrated into openFCST, an open-source FEM based fuel cell simulation framework. All the significant non-isothermal effects such as anisotropic heat transport, irreversible and reversible heating due to electrochemical reactions, ohmic heating, heat of sorption and thermal osmosis are accounted for in the model. Accurate experimental data for capillary pressure-saturation relationships are reformulated and incorporated into a liquid water saturation transport equation. The model showed, for the first time, that thermal osmosis effects account for up to 15% water-crossover in membrane, and the heat of sorption can be as large as protonic ohmic heating. The model could accurately predict the performance drops corresponding to mass transport limitations due to presence of liquid water and ohmic losses due to membrane dry-out.

Bitstream generation has traditionally been the single part of the FPGA design flow that has not been openly reproduced. This work enables bitstream generation for "limited" resources without reverse-engineering or violating End-User License Agreement terms. Two use cases in particular have motivated this work--embedded bitstream generation and fast bitstream generation for small changes in design--both of which are not feasible with the Xilinx's bitstream generation tool. The approach is to first define a set of primitives which can implement an arbitrary digital design and create a library of micro-bitstreams of the primitives. An input design is then mapped to the set of primitives and a bitstream for the design is generated by merging the corresponding micro-bitstreams. This work uses architectural primitives. Initial support is limited to the Virtex-5 and Virtex-7 family of FPGAs from Xilinx, but it can be extended to other Xilinx architectures. Nearly all routing resources in the device, as well as the most common logic resources are supported by this work.

Character rigs control characters in the traditional CG pipeline. This thesis examines the rig creation process and discusses several problems inherant in the traditional workflow--excessive time spent and a lack of uniformity, and proposes a software plugin which solves these issues. This thesis describes the creation of a tool for Blender 3D which automates the rigging process yet keeps the creativity and control in the hands of the user. A character rig designed by this tool will be fully functional--yet capable of being split into its component parts and reconstructed as the user determines. These body parts are individually scripted with the intent of maximizing reusability, and the code and rigs are distributed to the open source community for vetting. The final tool has been downloaded many times by the Blender community and has met with very postive responses.

The Xilinx Partial Reconfiguration tool kits have been instrumental for performing a wide variety of research on Xilinx FPGAs. These tool kits provide a methodology for creating rectangular partial reconfiguration modules that can be swapped in and out of a static baseline design with one or more PR slots. This thesis presents a new PR toolkit called OpenPR that, for starters, provides similar functionality to the Xilinx PR tool kits. The distinguishing feature of this toolkit is that it is being released as open source, and is intended to be customizable to the needs of researchers. OpenPR has been designed to be easy to use, extensible, portable, and compatible with a wide range of Xilinx software and devices. Aside from supporting the slot-based PR paradigm, OpenPR also provides a solid base for further research into partial reconfiguration and FPGA productivity oriented design tools.

Open Source Software (OSS) plays an increasingly important role in the social and economic development of countries, corporations, and individuals. Thus, it is important to understand the elements that affect OSS success. But, little is known on why certain projects succeed while others fail. Here, we draw on research in social network theory, and advance a conceptual model for project success in OSS systems. The current literature is dominated by a view that conceptualizes OSS projects as independent entities. This is inadequate in addressing issues related to the OSS, where each project is embedded in networks of relationships with other projects. Such relationship networks can be enduring and of strategic importance for OSS projects. We argue that the success of an OSS project can be more fully understood by adopting a network, rather than an independent, approach. Our model conceptualizes an OSS project as existing in a larger network of many related projects and identifies measurable network structures that are likely to distinguish successful OSS projects from less successful ones. We empirically evaluate the proposed model by analyzing panels of data from the world’s largest OSS development data repository, sourceforge.net.

An Open-Source, extensible spacecraft simulation and modeling (Open-SESSAME) framework was developed with the aim of providing to researchers the ability to quickly test satellite algorithms while allowing them the ability to view and extend the underlying code. The software is distributed under the GPL (General Public License) and the package's extensibility allows users to implement their own components into the libraries, investigate new algorithms, or tie in existing software or hardware components for algorithm and flight component testing. This thesis presents the purpose behind the development of the framework, the software design architecture and implementation, and a roadmap of the future for the software package.

A program has been developed to assist scientists in the biological sciences in the selection of primers for the polymerase chain reaction (PCR). Primer3 is a tool that can suggest primers based on requirements. The traditional query process encourages strict criteria be placed on primers to limit the number of primers suggested and to keep them within an ideal range. However, too often these restrictions are too strict and users must reformulate their queries and try again. Furthermore, there is little guidance on how to adjust the query. Now, broad criteria can be submitted to Primer3 and our program will allow the user to refine the criteria. The program differentiates those primers relating to new criteria using filters. The user can interactively adjust the filters on the suggested primers to see the effect. Primers not meeting the filter criteria are retained to help guide the user to adjust the filter. Interactions provided support dynamic exploration of the data set through allowing rapid, iterative, and reversible changes to filter criteria. Detailed information about selected primers is displayed in a text window for visual inspection and to facilitate a final choice. This solution has been developed using the open-source, cross-platform packages Tcl/Tk and VTK.