Applied Mathematics Commons^™

Improving Efficiency Of Rational Krylov Subspace Methods, Shengjie Xu

All Dissertations

This thesis studies two classes of numerical linear algebra problems, approximating the product of a function of a matrix with a vector, and solving the linear eigenvalue problem $Av=\lambda Bv$ for a small number of eigenvalues. These problems are solved by rational Krylov subspace methods (RKSM). We present several improvements in two directions: pole selection and applying inexact methods.

In Chapter 3, a flexible extended Krylov subspace method ($\mathcal{F}$-EKSM) is considered for numerical approximation of the action of a matrix function $f(A)$ to a vector $b$, where the function $f$ is of Markov type. $\mathcal{F}$-EKSM has the same framework as …

On Variants Of Sliding And Frank-Wolfe Type Methods And Their Applications In Video Co-Localization, Seyed Hamid Nazari

All Dissertations

In this dissertation, our main focus is to design and analyze first-order methods for computing approximate solutions to convex, smooth optimization problems over certain feasible sets. Specifically, our goal in this dissertation is to explore some variants of sliding and Frank-Wolfe (FW) type algorithms, analyze their convergence complexity, and examine their performance in numerical experiments. We achieve three accomplishments in our research results throughout this dissertation. First, we incorporate a linesearch technique to a well-known projection-free sliding algorithm, namely the conditional gradient sliding (CGS) method. Our proposed algorithm, called the conditional gradient sliding with linesearch (CGSls), does not require the …

Optimal First Order Methods For Reducing Gradient Norm In Unconstrained Convex Smooth Optimization, Yunheng Jiang

All Theses

In this thesis, we focus on convergence performance of first-order methods to compute an $\epsilon$-approximate solution of minimizing convex smooth function $f$ at the $N$-th iteration.

In our introduction of the above research question, we first introduce the gradient descent method with constant step size $h=1/L$. The gradient descent method has a $\mathcal{O}(L^2\|x_0-x^*\|^2/\epsilon)$ convergence with respect to $\|\nabla f(x_N)\|^2$. Next we introduce Nesterov’s accelerated gradient method, which has an $\mathcal{O}(L\|x_0-x^*\|\sqrt{1/\epsilon})$ complexity in terms of $\|\nabla f(x_N)\|^2$. The convergence performance of Nesterov’s accelerated gradient method is much better than that of the gradient descent method but still not optimal. We also …

Efficiency Of Homomorphic Encryption Schemes, Kyle Yates

All Theses

In 2009, Craig Gentry introduced the first fully homomorphic encryption scheme using bootstrapping. In the 13 years since, a large amount of research has gone into improving efficiency of homomorphic encryption schemes. This includes implementing leveled homomorphic encryption schemes for practical use, which are schemes that allow for some predetermined amount of additions and multiplications that can be performed on ciphertexts. These leveled schemes have been found to be very efficient in practice. In this thesis, we will discuss the efficiency of various homomorphic encryption schemes. In particular, we will see how to improve sizes of parameter choices in homomorphic …

A Conservative Numerical Scheme For The Multilayer Shallow Water Equations, Evan Butterworth

All Theses

An energy-conserving numerical scheme is developed for the multilayer shallow water equations (SWE’s). The scheme is derived through the Hamiltonian formulation of the inviscid shallow water flows related to the vorticity-divergence variables. Through the employment of the skew-symmetric Poisson bracket, the continuous system for the multilayer SWE’s is shown to preserve an infinite number of quantities, most notably the energy and enstrophy. An energy-preserving numerical scheme is then developed through the careful discretization of the Hamiltonian and the Poisson bracket, ensuring the skew-symmetry of the latter. This serves as the groundwork for developing additional schemes that preserve other conservation properties …

Managing Risk For Power System Operations And Planning: Applications Of Conditional Value-At-Risk And Uncertainty Quantification To Optimal Power Flow And Distributed Energy Resources Investment, Thanh To

All Dissertations

Renewable energy sources are indispensable components of sustainable electrical systems that reduce human dependence on fossil fuels. However, due to their intermittent nature, there are issues that need to be addressed to ensure the security and resiliency of these power systems. This dissertation formulates several practical problems, from an optimization perspective, stemming from the increasing penetration of intermittent renewable energy to power systems. A number of Optimal Power Flow (OPF) formulations are investigated and new formulations are proposed to control both operations and planning risks by utilizing the Conditional Value–at–Risk (CVaR) measure. Our formulations provide system operators and investors analysis …

Advancements In Gaussian Process Learning For Uncertainty Quantification, John C. Nicholson

All Dissertations

Gaussian processes are among the most useful tools in modeling continuous processes in machine learning and statistics. The research presented provides advancements in uncertainty quantification using Gaussian processes from two distinct perspectives. The first provides a more fundamental means of constructing Gaussian processes which take on arbitrary linear operator constraints in much more general framework than its predecessors, and the other from the perspective of calibration of state-aware parameters in computer models. If the value of a process is known at a finite collection of points, one may use Gaussian processes to construct a surface which interpolates these values to …

An Algorithm For Biobjective Mixed Integer Quadratic Programs, Pubudu Jayasekara Merenchige

All Dissertations

Multiobjective quadratic programs (MOQPs) are appealing since convex quadratic programs have elegant mathematical properties and model important applications. Adding mixed-integer variables extends their applicability while the resulting programs become global optimization problems. Thus, in this work, we develop a branch and bound (BB) algorithm for solving biobjective mixed-integer quadratic programs (BOMIQPs). An algorithm of this type does not exist in the literature.

The algorithm relies on five fundamental components of the BB scheme: calculating an initial set of efficient solutions with associated Pareto points, solving node problems, fathoming, branching, and set dominance. Considering the properties of the Pareto set of …

A Parallelized And Layered Model For The Shallow-Water Equations, Alexander Stevens

All Theses

An energy- and enstrophy-conserving and optimally-dispersive numerical scheme for the shallow- water equations is accelerated through implementation in the GPU environment. Previous research showed the viability of the numerical scheme under standard shallow-water test cases, but was limited in applications by computation time constraints. We overcome these limitations by paral- lelizing the numerical computation in the GPU environment. We also extend the capabilities of the implementation to support not just a single shallow-water layer, but multiple. These improvements significantly expand the range of tests that can be used to exercise the model, and enable better understanding of the power of …

Numerical Decoding, Johnson-Lindenstrauss Transforms, And Linear Codes, Yue Mao

All Dissertations

Many computational problems are related to the model y = Ax + e, including compressive sensing, coding theory, dimensionality reduction, etc. The related algorithms are extremely useful in practical applications for high performance computing, for example, digital communications, biological imaging and data streaming, etc. This thesis studies two important problems. One problem is related to efficient decoding for Reed-Solomon codes over complex numbers. In this case, A and y are given, and the goal is to find an efficient stable algorithm to compute x. This is related to magnetic resonance imaging (MRI). The other problem is related to fast algorithms …

Computational Exploration Of Chaotic Dynamics With An Associated Biological System, Akshay Galande

All Theses

Study of microbial populations has always been topic of interest for researchers. This is because microorganisms have been of instrumental use in the various studies related to population dynamics, artificial bio-fuels etc. Comparatively short lifespan and availability are two big advantages they have which make them suitable for aforementioned studies. Their population dynamic helps us understand evolution. A lot can be revealed about resource consumption of a system by comparing it to the similar system where bacteria play the role of different factors in the system. Also, study of population dynamics of bacteria can reveal necessary initial conditions for the …

Computational Bases For Hdiv, Alistair Bentley

All Theses

The \(H_{div}\) vector space arises in a number of mixed method formulations, particularly in fluid flow through a porous medium. First we present a Lagrangian computational basis for the Raviert-Thomas (\(RT\)) and Brezzi-Douglas-Marini (\(BDM\)) approximation subspaces of \(H_{div}\) in \(\mathbb{R}^{3}\). Second, we offer three solutions to a numerical problem that arises from the Piola mapping when \(RT\) and \(BDM\) elements are used in practice.

Homomorphic Encryption And The Approximate Gcd Problem, Nathanael Black

All Dissertations

With the advent of cloud computing, everyone from Fortune 500 businesses to personal consumers to the US government is storing massive amounts of sensitive data in service centers that may not be trustworthy. It is of vital importance to leverage the benefits of storing data in the cloud while simultaneously ensuring the privacy of the data. Homomorphic encryption allows one to securely delegate the processing of private data. As such, it has managed to hit the sweet spot of academic interest and industry demand. Though the concept was proposed in the 1970s, no cryptosystem realizing this goal existed until Craig …

Applied Statistics In Environmental Monitoring: Case Studies And Analysis For The Michigan Bald Eagle Biosentinel Program, Katherine Leith

All Dissertations

The bald eagle (Haliaeetus leucocephalus) is an extensively researched tertiary predator. Its life history and the impact of various stressors on its reproductive outcomes have been documented in many studies, and over many years. Furthermore, the bald eagle population recovery in Michigan has been closely monitored since the 1960s, as it has continued to recover from a contaminant-induced bottleneck. Because of its position at the top of the aquatic food web and the large body of ethological knowledge, the bald eagle has become a sentinel species for the Michigan aquatic ecosystem. In April 1999, the Michigan Department of Environmental Qualtity, …

Grobner Bases: Degree Bounds And Generic Ideals, Juliane Golubinski Capaverde

All Dissertations

In this thesis, we study two problems related to Gröbner basis theory: degree bounds for general ideals and Gröbner bases structure for generic ideals. We start by giving an introduction to Gröbner bases and their basic properties and presenting a recent algorithm by Gao, Volny and Wang. Next, we survey degree bounds for the ideal membership problem, the effective Nullstellensatz, and polynomials in minimal Gröbner bases. We present general upper bounds, and bounds for several classes of special ideals. We provide classical examples showing some of these bounds cannot be improved in general. We present a comprehensive study of a …

On Numerical Algorithms For Fluid Flow Regularization Models, Abigail Bowers

All Dissertations

This thesis studies regularization models as a way to approximate a flow simulation at a lower computational cost. The Leray model is more easily computed than the Navier-Stokes equations (NSE), and it is more computationally attractive than the NS-α regularization because it admits a natural linearization which decouples the mass/momentum system and the filter system, allowing for efficient and stable computations. A major disadvantage of the Leray model lies in its inaccuracy. Thus, we study herein several methods to improve the accuracy of the model, while still retaining many of its attractive properties. This thesis is arranged as follows. Chapter …

Level Stripping Of Genus 2 Siegel Modular Forms, Rodney Keaton

All Dissertations

In this Dissertation we consider stripping primes from the level of genus 2 cuspidal Siegel eigenforms. Specifically, given an eigenform of level Nlr which satisfies certain mild conditions, where l is a prime not dividing N, we construct an eigenform of level N which is congruent to our original form. To obtain our results, we use explicit constructions of Eisenstein series and theta functions to adapt ideas from a level stripping result on elliptic modular forms. Furthermore, we give applications of this result to Galois representations and provide evidence for an analog of Serre's conjecture in the genus 2 case.

Convergence Of A Reinforcement Learning Algorithm In Continuous Domains, Stephen Carden

All Dissertations

In the field of Reinforcement Learning, Markov Decision Processes with a finite number of states and actions have been well studied, and there exist algorithms capable of producing a sequence of policies which converge to an optimal policy with probability one. Convergence guarantees for problems with continuous states also exist. Until recently, no online algorithm for continuous states and continuous actions has been proven to produce optimal policies. This Dissertation contains the results of research into reinforcement learning algorithms for problems in which both the state and action spaces are continuous. The problems to be solved are introduced formally as …

Improved Mixed-Integer Models Of A Two-Dimensional Cutting Stock Problem, William Lassiter

All Theses

This paper is concerned with a family of two-dimensional cutting stock problems that seeks to cut rectangular regions from a finite collection of sheets in such a manner that the minimum number of sheets is used. A fixed number of rectangles are to be cut, with each rectangle having a known length and width. All sheets are rectangular, and have the same dimension. We review two known mixed-integer mathematical formulations, and then provide new representations that both economize on the number of discrete variables and tighten the continuous relaxations. A key consideration that arises repeatedly in all models is the …

The Intelligent Driver Model: Analysis And Application To Adaptive Cruise Control, Rachel Malinauskas

All Theses

There are a large number of models that can be used to describe traffic flow. Although some were initially theoretically derived, there are many that were constructed with utility alone in mind. The Intelligent Driver Model (IDM) is a microscopic model that can be used to examine traffic behavior on an individual level with emphasis on the relation to an ahead vehicle. One application for this model is that it is easily molded to performing the operations for an Adaptive Cruise Control (ACC) system. Although it is clear that the IDM holds a number of convenient properties, like easily interpreted …

Approximation In Multiobjective Optimization With Applications, Lakmali Weerasena

All Dissertations

Over the last couple of decades, the field of multiobjective optimization has received much attention in solving real-life optimization problems in science, engineering, economics and other fields where optimal decisions need to be made in the presence of trade-offs between two or more conflicting objective functions. The conflicting nature of objective functions implies a solution set for a multiobjective optimization problem. Obtaining this set is difficult for many reasons, and a variety of approaches for approximating it either partially or entirely have been proposed.

In response to the growing interest in approximation, this research investigates developing a theory and methodology …

Mathematical Optimization For Engineering Design Problems, Brian Dandurand

All Dissertations

Applications in engineering design and the material sciences motivate the development of optimization theory in a manner that additionally draws from other branches of mathematics including the functional, complex, and numerical analyses.
The first contribution, motivated by an automotive design application, extends multiobjective optimization theory under the assumption that the problem information is not available in its entirety to a single decision maker as traditionally assumed in the multiobjective optimization literature. Rather, the problem information and the design control are distributed among different decision makers. This requirement appears in the design of an automotive system whose subsystem components themselves correspond …

On The Cuspidality Of Maass-Gritsenko And Mixed Level Lifts, Dania Zantout

All Dissertations

your words

Advancements In Finite Element Methods For Newtonian And Non-Newtonian Flows, Keith Galvin

All Dissertations

This dissertation studies two important problems in the mathematics of computational fluid dynamics. The first problem concerns the accurate and efficient simulation of incompressible, viscous Newtonian flows, described by the Navier-Stokes equations. A direct numerical simulation of these types of flows is, in most cases, not computationally feasible. Hence, the first half of this work studies two separate types of models designed to more accurately and efficient simulate these flows. The second half focuses on the defective boundary problem for non-Newtonian flows. Non-Newtonian flows are generally governed by more complex modeling equations, and the lack of standard Dirichlet or Neumann …

Secret Sharing And Network Coding, Fiona Knoll

All Theses

In this thesis, we consider secret sharing schemes and network coding. Both of these fields are vital in today's age as secret sharing schemes are currently being implemented by government agencies and private companies, and as network coding is continuously being used for IP networks. We begin with a brief overview of linear codes. Next, we examine van Dijk's approach to realize an access structure using a linear secret sharing scheme; then we focus on a much simpler approach by Tang, Gao, and Chen. We show how this method can be used to find an optimal linear secret sharing scheme …

Polyhedral Approximations Of Quadratic Semi-Assignment Problems, Disjunctive Programs, And Base-2 Expansions Of Integer Variables, Frank Muldoon

All Dissertations

This research is concerned with developing improved representations for special families of mixed-discrete programming problems. Such problems can typically be modeled using different mathematical forms, and the representation employed can greatly influence the problem's ability to be solved. Generally speaking, it is desired to obtain mixed 0-1 linear forms whose continuous relaxations provide tight polyhedral outer-approximations to the convex hulls of feasible solutions. This dissertation makes contributions to three distinct problems, providing new forms that improve upon published works.
The first emphasis is on devising solution procedures for the classical quadratic semi-assignment problem(QSAP), which is an NP-hard 0-1 quadratic program. …

Convex Hull Characterization Of Special Polytopes In N-Ary Variables, Ruobing Shen

All Theses

This paper characterizes the convex hull of the set of n-ary vectors that are lexicographically less than or equal to a given such vector. A polynomial number of facets is shown to be sufficient to describe the convex hull. These facets generalize the family of cover inequalities for the binary case. They allow for advances relative to both the modeling of integer variables using base-n expansions, and the solving of n-ary knapsack problems with weakly super-decreasing coefficients.

Sensitivity Analysis In Magnetic Resonance Elastography And A Local Wavelength Reconstruction Based On Wave Direction, Christopher Gillam

All Dissertations

or the detection of early stage cancer. MRE utilizes interior data for its inverse problems, which greatly reduces the ill-posedness from which most traditional inverse problems suffer.
In this thesis, we first establish a sensitivity analysis for viscoelastic scalar medium with complex wave number and compare it with the purely elastic case. Also we estimate the smallest detectable inclusion for breast and liver, which is about twice larger than using the purely elastic model. We also found the existence of optimal frequency (50 Hz) that maximizes the detectability when the Voigt model is used.
Second, we propose a local wavelength …

Local Polynomial Regression With Application To Sea Surface Temperatures, Michael Finney

All Theses

Our problem involves methods for determining the times of a maximum or minimum for a general mean function in time series data. The methods explored here involve polynomial smoothing. In theory, the methods calculate a general number of derivatives of the estimated polynomial. Using these techniques, we wish to find a balance between error, variance, and complexity and apply it to a time series of sea surface temperatures. We will first explore the theory behind the method and then find a way to optimally apply it to our data.

Robust Parameter Estimation In The Weibull And The Birnbaum-Saunders Distribution, Jing Zhao

All Theses

This paper concerns robust parameter estimation of the two-parameter Weibull distribution and the two-parameter Birnbaum-Saunders distribution. We use the proposed method to estimate the distribution parameters from (i) complete samples with and without contamination (ii) type-II censoring samples, in both distributions. Also, we consider the maximum likelihood estimation and graphical methods to compare the maximum likelihood estimation and graphical method with the proposed method based on quantile. We find the advantages and disadvantages for those three different methods.