Problems in Algebraic Vision

Thesis (Ph.D.)--University of Washington, 2016-08

A finite-volume computational mechanics framework for multi-physics coupled fluid-stress problems

Abstract not available

Finite volume methods for multiphysics problems

Abstract not available

Comparison of three algorithms for nonlinear metal cutting problems

This paper compares three alternative numerical algorithms applied to a nonlinear metal cutting problem. One algorithm is based on an explicit method and the other two are implicit. Domain decomposition (DD) is used to break the original domain into subdomains, each containing a properly connected, well-formulated and continuous subproblem. The serial version of the explicit algorithm is implemented in FORTRAN and its parallel version uses MPI (Message Passing Interface) calls. One implicit algorithm is implemented by coupling the state-of-the-art PETSc (Portable, Extensible Toolkit for Scientific Computation) software with in-house software in order to solve the subproblems. The second implicit algorithm is implemented completely within PETSc. PETSc uses MPI as the underlying communication library. Finally, a 2D example is used to test the algorithms and various comparisons are made.

Riemann-Hilbert problems for poly-Hardy space on the unit ball

In this paper, we focus on a Riemann–Hilbert boundary value problem (BVP) with a constant coefficients for the poly-Hardy space on the real unit ball in higher dimensions. We first discuss the boundary behaviour of functions in the poly-Hardy class. Then we construct the Schwarz kernel and the higher order Schwarz operator to study Riemann–Hilbert BVPs over the unit ball for the poly- Hardy class. Finally, we obtain explicit integral expressions for their solutions. As a special case, monogenic signals as elements in the Hardy space over the unit sphere will be reconstructed in the case of boundary data given in terms of functions having values in a Clifford subalgebra. Such monogenic signals represent the generalization of analytic signals as elements of the Hardy space over the unit circle of the complex plane.

HPC management and engineering in the hybrid cloud

The evolution and maturation of Cloud Computing created an opportunity for the emergence of new Cloud applications. High-performance Computing, a complex problem solving class, arises as a new business consumer by taking advantage of the Cloud premises and leaving the expensive datacenter management and difficult grid development. Standing on an advanced maturing phase, today’s Cloud discarded many of its drawbacks, becoming more and more efficient and widespread. Performance enhancements, prices drops due to massification and customizable services on demand triggered an emphasized attention from other markets. HPC, regardless of being a very well established field, traditionally has a narrow frontier concerning its deployment and runs on dedicated datacenters or large grid computing. The problem with common placement is mainly the initial cost and the inability to fully use resources which not all research labs can afford. The main objective of this work was to investigate new technical solutions to allow the deployment of HPC applications on the Cloud, with particular emphasis on the private on-premise resources – the lower end of the chain which reduces costs. The work includes many experiments and analysis to identify obstacles and technology limitations. The feasibility of the objective was tested with new modeling, architecture and several applications migration. The final application integrates a simplified incorporation of both public and private Cloud resources, as well as HPC applications scheduling, deployment and management. It uses a well-defined user role strategy, based on federated authentication and a seamless procedure to daily usage with balanced low cost and performance.

Tissue engineering: technological advances to improve its applications in reconstructive surgery

Background. Tremendous advances in biomaterials science and nanotechnologies, together with thorough research on stem cells, have recently promoted an intriguing development of regenerative medicine/tissue engineering. The nanotechnology represents a wide interdisciplinary field that implies the manipulation of different materials at nanometer level to achieve the creation of constructs that mimic the nanoscale-based architecture of native tissues. Aim. The purpose of this article is to highlight the significant new knowledges regarding this matter. Emerging acquisitions. To widen the range of scaffold materials resort has been carried out to either recombinant DNA technology-generated materials, such as a collagen-like protein, or the incorporation of bioactive molecules, such as RDG (arginine-glycine-aspartic acid), into synthetic products. Both the bottom-up and the top-down fabrication approaches may be properly used to respectively obtain sopramolecular architectures or, instead, micro-/nanostructures to incorporate them within a preexisting complex scaffold construct. Computer-aided design/manufacturing (CAD/CAM) scaffold technique allows to achieve patient-tailored organs. Stem cells, because of their peculiar properties - ability to proliferate, self-renew and specific cell-lineage differentiate under appropriate conditions - represent an attractive source for intriguing tissue engineering/regenerative medicine applications. Future research activities. New developments in the realization of different organs tissue engineering will depend on further progress of both the science of nanoscale-based materials and the knowledge of stem cell biology. Moreover the in vivo tissue engineering appears to be the logical step of the current research.

Anomaly Detection in Noisy Images

Finding rare events in multidimensional data is an important detection problem that has applications in many fields, such as risk estimation in insurance industry, finance, flood prediction, medical diagnosis, quality assurance, security, or safety in transportation. The occurrence of such anomalies is so infrequent that there is usually not enough training data to learn an accurate statistical model of the anomaly class. In some cases, such events may have never been observed, so the only information that is available is a set of normal samples and an assumed pairwise similarity function. Such metric may only be known up to a certain number of unspecified parameters, which would either need to be learned from training data, or fixed by a domain expert. Sometimes, the anomalous condition may be formulated algebraically, such as a measure exceeding a predefined threshold, but nuisance variables may complicate the estimation of such a measure. Change detection methods used in time series analysis are not easily extendable to the multidimensional case, where discontinuities are not localized to a single point. On the other hand, in higher dimensions, data exhibits more complex interdependencies, and there is redundancy that could be exploited to adaptively model the normal data. In the first part of this dissertation, we review the theoretical framework for anomaly detection in images and previous anomaly detection work done in the context of crack detection and detection of anomalous components in railway tracks. In the second part, we propose new anomaly detection algorithms. The fact that curvilinear discontinuities in images are sparse with respect to the frame of shearlets, allows us to pose this anomaly detection problem as basis pursuit optimization. Therefore, we pose the problem of detecting curvilinear anomalies in noisy textured images as a blind source separation problem under sparsity constraints, and propose an iterative shrinkage algorithm to solve it. Taking advantage of the parallel nature of this algorithm, we describe how this method can be accelerated using graphical processing units (GPU). Then, we propose a new method for finding defective components on railway tracks using cameras mounted on a train. We describe how to extract features and use a combination of classifiers to solve this problem. Then, we scale anomaly detection to bigger datasets with complex interdependencies. We show that the anomaly detection problem naturally fits in the multitask learning framework. The first task consists of learning a compact representation of the good samples, while the second task consists of learning the anomaly detector. Using deep convolutional neural networks, we show that it is possible to train a deep model with a limited number of anomalous examples. In sequential detection problems, the presence of time-variant nuisance parameters affect the detection performance. In the last part of this dissertation, we present a method for adaptively estimating the threshold of sequential detectors using Extreme Value Theory on a Bayesian framework. Finally, conclusions on the results obtained are provided, followed by a discussion of possible future work.

DIRECT NUMERICAL SIMULATION OF INCOMPRESSIBLE MULTIPHASE FLOW WITH PHASE CHANGE

Phase change problems arise in many practical applications such as air-conditioning and refrigeration, thermal energy storage systems and thermal management of electronic devices. The physical phenomenon in such applications are complex and are often difficult to be studied in detail with the help of only experimental techniques. The efforts to improve computational techniques for analyzing two-phase flow problems with phase change are therefore gaining momentum. The development of numerical methods for multiphase flow has been motivated generally by the need to account more accurately for (a) large topological changes such as phase breakup and merging, (b) sharp representation of the interface and its discontinuous properties and (c) accurate and mass conserving motion of the interface. In addition to these considerations, numerical simulation of multiphase flow with phase change introduces additional challenges related to discontinuities in the velocity and the temperature fields. Moreover, the velocity field is no longer divergence free. For phase change problems, the focus of developmental efforts has thus been on numerically attaining a proper conservation of energy across the interface in addition to the accurate treatment of fluxes of mass and momentum conservation as well as the associated interface advection. Among the initial efforts related to the simulation of bubble growth in film boiling applications the work in \cite{Welch1995} was based on the interface tracking method using a moving unstructured mesh. That study considered moderate interfacial deformations. A similar problem was subsequently studied using moving, boundary fitted grids \cite{Son1997}, again for regimes of relatively small topological changes. A hybrid interface tracking method with a moving interface grid overlapping a static Eulerian grid was developed \cite{Juric1998} for the computation of a range of phase change problems including, three-dimensional film boiling \cite{esmaeeli2004computations}, multimode two-dimensional pool boiling \cite{Esmaeeli2004} and film boiling on horizontal cylinders \cite{Esmaeeli2004a}. The handling of interface merging and pinch off however remains a challenge with methods that explicitly track the interface. As large topological changes are crucial for phase change problems, attention has turned in recent years to front capturing methods utilizing implicit interfaces that are more effective in treating complex interface deformations. The VOF (Volume of Fluid) method was adopted in \cite{Welch2000} to simulate the one-dimensional Stefan problem and the two-dimensional film boiling problem. The approach employed a specific model for mass transfer across the interface involving a mass source term within cells containing the interface. This VOF based approach was further coupled with the level set method in \cite{Son1998}, employing a smeared-out Heaviside function to avoid the numerical instability related to the source term. The coupled level set, volume of fluid method and the diffused interface approach was used for film boiling with water and R134a at the near critical pressure condition \cite{Tomar2005}. The effect of superheat and saturation pressure on the frequency of bubble formation were analyzed with this approach. The work in \cite{Gibou2007} used the ghost fluid and the level set methods for phase change simulations. A similar approach was adopted in \cite{Son2008} to study various boiling problems including three-dimensional film boiling on a horizontal cylinder, nucleate boiling in microcavity \cite{lee2010numerical} and flow boiling in a finned microchannel \cite{lee2012direct}. The work in \cite{tanguy2007level} also used the ghost fluid method and proposed an improved algorithm based on enforcing continuity and divergence-free condition for the extended velocity field. The work in \cite{sato2013sharp} employed a multiphase model based on volume fraction with interface sharpening scheme and derived a phase change model based on local interface area and mass flux. Among the front capturing methods, sharp interface methods have been found to be particularly effective both for implementing sharp jumps and for resolving the interfacial velocity field. However, sharp velocity jumps render the solution susceptible to erroneous oscillations in pressure and also lead to spurious interface velocities. To implement phase change, the work in \cite{Hardt2008} employed point mass source terms derived from a physical basis for the evaporating mass flux. To avoid numerical instability, the authors smeared the mass source by solving a pseudo time-step diffusion equation. This measure however led to mass conservation issues due to non-symmetric integration over the distributed mass source region. The problem of spurious pressure oscillations related to point mass sources was also investigated by \cite{Schlottke2008}. Although their method is based on the VOF, the large pressure peaks associated with sharp mass source was observed to be similar to that for the interface tracking method. Such spurious fluctuation in pressure are essentially undesirable because the effect is globally transmitted in incompressible flow. Hence, the pressure field formation due to phase change need to be implemented with greater accuracy than is reported in current literature. The accuracy of interface advection in the presence of interfacial mass flux (mass flux conservation) has been discussed in \cite{tanguy2007level,tanguy2014benchmarks}. The authors found that the method of extending one phase velocity to entire domain suggested by Nguyen et al. in \cite{nguyen2001boundary} suffers from a lack of mass flux conservation when the density difference is high. To improve the solution, the authors impose a divergence-free condition for the extended velocity field by solving a constant coefficient Poisson equation. The approach has shown good results with enclosed bubble or droplet but is not general for more complex flow and requires additional solution of the linear system of equations. In current thesis, an improved approach that addresses both the numerical oscillation of pressure and the spurious interface velocity field is presented by featuring (i) continuous velocity and density fields within a thin interfacial region and (ii) temporal velocity correction steps to avoid unphysical pressure source term. Also I propose a general (iii) mass flux projection correction for improved mass flux conservation. The pressure and the temperature gradient jump condition are treated sharply. A series of one-dimensional and two-dimensional problems are solved to verify the performance of the new algorithm. Two-dimensional and cylindrical film boiling problems are also demonstrated and show good qualitative agreement with the experimental observations and heat transfer correlations. Finally, a study on Taylor bubble flow with heat transfer and phase change in a small vertical tube in axisymmetric coordinates is carried out using the new multiphase, phase change method.

Exact Solution of Bayes and Minimax Change-Detection Problems

The challenge of detecting a change in the distribution of data is a sequential decision problem that is relevant to many engineering solutions, including quality control and machine and process monitoring. This dissertation develops techniques for exact solution of change-detection problems with discrete time and discrete observations. Change-detection problems are classified as Bayes or minimax based on the availability of information on the change-time distribution. A Bayes optimal solution uses prior information about the distribution of the change time to minimize the expected cost, whereas a minimax optimal solution minimizes the cost under the worst-case change-time distribution. Both types of problems are addressed. The most important result of the dissertation is the development of a polynomial-time algorithm for the solution of important classes of Markov Bayes change-detection problems. Existing techniques for epsilon-exact solution of partially observable Markov decision processes have complexity exponential in the number of observation symbols. A new algorithm, called constellation induction, exploits the concavity and Lipschitz continuity of the value function, and has complexity polynomial in the number of observation symbols. It is shown that change-detection problems with a geometric change-time distribution and identically- and independently-distributed observations before and after the change are solvable in polynomial time. Also, change-detection problems on hidden Markov models with a fixed number of recurrent states are solvable in polynomial time. A detailed implementation and analysis of the constellation-induction algorithm are provided. Exact solution methods are also established for several types of minimax change-detection problems. Finite-horizon problems with arbitrary observation distributions are modeled as extensive-form games and solved using linear programs. Infinite-horizon problems with linear penalty for detection delay and identically- and independently-distributed observations can be solved in polynomial time via epsilon-optimal parameterization of a cumulative-sum procedure. Finally, the properties of policies for change-detection problems are described and analyzed. Simple classes of formal languages are shown to be sufficient for epsilon-exact solution of change-detection problems, and methods for finding minimally sized policy representations are described.

FRAMEWORK SYNTHESIS FOR SYMBOLIC EXECUTION OF EVENT-DRIVEN FRAMEWORKS

Symbolic execution is a powerful program analysis technique, but it is very challenging to apply to programs built using event-driven frameworks, such as Android. The main reason is that the framework code itself is too complex to symbolically execute. The standard solution is to manually create a framework model that is simpler and more amenable to symbolic execution. However, developing and maintaining such a model by hand is difficult and error-prone. We claim that we can leverage program synthesis to introduce a high-degree of automation to the process of framework modeling. To support this thesis, we present three pieces of work. First, we introduced SymDroid, a symbolic executor for Android. While Android apps are written in Java, they are compiled to Dalvik bytecode format. Instead of analyzing an app’s Java source, which may not be available, or decompiling from Dalvik back to Java, which requires significant engineering effort and introduces yet another source of potential bugs in an analysis, SymDroid works directly on Dalvik bytecode. Second, we introduced Pasket, a new system that takes a first step toward automatically generating Java framework models to support symbolic execution. Pasket takes as input the framework API and tutorial programs that exercise the framework. From these artifacts and Pasket's internal knowledge of design patterns, Pasket synthesizes an executable framework model by instantiating design patterns, such that the behavior of a synthesized model on the tutorial programs matches that of the original framework. Lastly, in order to scale program synthesis to framework models, we devised adaptive concretization, a novel program synthesis algorithm that combines the best of the two major synthesis strategies: symbolic search, i.e., using SAT or SMT solvers, and explicit search, e.g., stochastic enumeration of possible solutions. Adaptive concretization parallelizes multiple sub-synthesis problems by partially concretizing highly influential unknowns in the original synthesis problem. Thanks to adaptive concretization, Pasket can generate a large-scale model, e.g., thousands lines of code. In addition, we have used an Android model synthesized by Pasket and found that the model is sufficient to allow SymDroid to execute a range of apps.

Evolution feature oriented model driven product line engineering approach for synergistic and dynamic service evolution in clouds: A04BPEL3.0 proposal.

Model Driven based approach for Service Evolution in Clouds will mainly focus on the reusable evolution patterns' advantage to solve evolution problems. During the process, evolution pattern will be driven by MDA models to pattern aspects. Weaving the aspects into service based process by using Aspect-Oriented extended BPEL engine at runtime will be the dynamic feature of the evolution.

EXCITON ENGINEERING THROUGH TUNALBLE FLUORESCENT QUANTUM DEFECTS

This thesis demonstrates exciton engineering in semiconducting single-walled carbon nanotubes through tunable fluorescent quantum defects. By introducing different functional moieties on the sp2 lattice of carbon nanotubes, the nanotube photoluminescence is systematically tuned over 68 meV in the second near-infrared window. This new class of quantum emitters is enabled by a new chemistry that allows covalent attachment of alkyl/aryl functional groups from their iodide precursors in aqueous solution. Using aminoaryl quantum defects, we show that the pH and temperature of complex fluids can be optically measured through defect photoluminescence that encodes the local environment information. Furthermore, defect-bound trions, which are electron-hole-electron tri-carrier quasi-particles, are observed in alkylated single-walled carbon nanotubes at room temperature with surprisingly high photoluminescence brightness. Collectively, the emission from defect-bound excitons and trions in (6,5)-single walled carbon nanotubes is 18-fold brighter than that of the native exciton. These findings pave the way to chemical tailoring of the electronic and optical properties of carbon nanostructures with fluorescent quantum defects and may find applications in optoelectronics and bioimaging.

Case-Based Reasoning as a Heuristic Selector in a Hyper-Heuristic for Course Timetabling Problems

This paper studies Knowledge Discovery (KD) using Tabu Search and Hill Climbing within Case-Based Reasoning (CBR) as a hyper-heuristic method for course timetabling problems. The aim of the hyper-heuristic is to choose the best heuristic(s) for given timetabling problems according to the knowledge stored in the case base. KD in CBR is a 2-stage iterative process on both case representation and the case base. Experimental results are analysed and related research issues for future work are discussed.

Ecology, physiology and performance in high-rate anaerobic digestion

The design demands on water and sanitation engineers are rapidly changing. The global population is set to rise from 7 billion to 10 billion by 2083. Urbanisation in developing regions is increasing at such a rate that a predicted 56% of the global population will live in an urban setting by 2025. Compounding these problems, the global water and energy crises are impacting the Global North and South alike. High-rate anaerobic digestion offers a low-cost, low-energy treatment alternative to the energy intensive aerobic technologies used today. Widespread implementation however is hindered by the lack of capacity to engineer high-rate anaerobic digestion for the treatment of complex wastes such as sewage. This thesis utilises the Expanded Granular Sludge Bed bioreactor (EGSB) as a model system in which to study the ecology, physiology and performance of high-rate anaerobic digestion of complex wastes. The impacts of a range of engineered parameters including reactor geometry, wastewater type, operating temperature and organic loading rate are systematically investigated using lab-scale EGSB bioreactors. Next generation sequencing of 16S amplicons is utilised as a means of monitoring microbial ecology. Microbial community physiology is monitored by means of specific methanogenic activity testing and a range of physical and chemical methods are applied to assess reactor performance. Finally, the limit state approach is trialled as a method for testing the EGSB and is proposed as a standard method for biotechnology testing enabling improved process control at full-scale. The arising data is assessed both qualitatively and quantitatively. Lab-scale reactor design is demonstrated to significantly influence the spatial distribution of the underlying ecology and community physiology in lab-scale reactors, a vital finding for both researchers and full-scale plant operators responsible for monitoring EGSB reactors. Recurrent trends in the data indicate that hydrogenotrophic methanogenesis dominates in high-rate anaerobic digestion at both full- and lab-scale when subject to engineered or operational stresses including low-temperature and variable feeding regimes. This is of relevance for those seeking to define new directions in fundamental understanding of syntrophic and competitive relations in methanogenic communities and also to design engineers in determining operating parameters for full-scale digesters. The adoption of the limit state approach enabled identification of biological indicators providing early warning of failure under high-solids loading, a vital insight for those currently working empirically towards the development of new biotechnologies at lab-scale.