Applied Mathematics Commons^™

Algebraic, Computational, And Data-Driven Methods For Control-Theoretic Analysis And Learning Of Ensemble Systems, Wei Miao

McKelvey School of Engineering Graduate Student Theses & Dissertations

In this thesis, we study a class of problems involving a population of dynamical systems under a common control signal, namely, ensemble systems, through both control-theoretic and data-driven perspectives. These problems are stemmed from the growing need to understand and manipulate large collections of dynamical systems in emerging scientific areas such as quantum control, neuroscience, and magnetic resonance imaging. We examine fundamental control-theoretic properties such as ensemble controllability of ensemble systems and ensemble reachability of ensemble states, and propose ensemble control design approaches to devise control signals that steer ensemble systems to desired profiles. We show that these control-theoretic properties …

Lower Bounds On Betti Numbers, Adam Boocher, Eloisa Grifo

Department of Mathematics: Faculty Publications

We survey recent results on bounds for Betti numbers of modules over polynomial rings, with an emphasis on lower bounds. Along the way, we give a gentle introduction to free resolutions and Betti numbers, and discuss some of the reasons why one would study these.

Empirical Fitting Of Periodically Repeating Environmental Data, Pavel Bělík, Andrew Hotchkiss, Brandon Perez, John Zobitz

Spora: A Journal of Biomathematics

We extend and generalize an approach to conduct fitting models of periodically repeating data. Our method first detrends the data from a baseline function and then fits the data to a periodic (trigonometric, polynomial, or piecewise linear) function. The polynomial and piecewise linear functions are developed from assumptions of continuity and differentiability across each time period. We apply this approach to different datasets in the environmental sciences in addition to a synthetic dataset. Overall the polynomial and piecewise linear approaches developed here performed as good (or better) compared to the trigonometric approach when evaluated using statistical measures (R^{2 …}

Sparse Domination Of The Martingale Transform, Michael Scott Kutzler

Mathematics & Statistics ETDs

Linear operators are of huge importance in modern harmonic analysis. Many operators can be dominated by finitely many sparse operators. The main result in this thesis is showing a toy operator, namely the Martingale Transform is dominated by a single sparse operator. Sparse operators are based on a sparse family which is simply a subset of a dyadic grid. We also show the A2 conjecture for the Martingale Transform which follows from the sparse domination of the Martingale Transform and the A2 conjecture for sparse operators.

.

Growth-Profile Configuration For Specific Deformations Of Tubular Organs: A Study Of Growth-Induced Thinning And Dilation Of The Human Cervix, Kun Gou, Seungik Baek, Marvin M.F. Lutnesky, Hai-Chao Han

Mathematics Faculty Publications (Archived)

Growth is a significant factor that results in deformations of tubular organs, and particular deformations associated with growth enable tubular organs to perform certain physiological functions. Configuring growth profiles that achieve particular deformation patterns is critical for analyzing potential pathological conditions and for developing corresponding clinical treatments for tubular organ dysfunctions. However, deformation-targeted growth is rarely studied. In this article, the human cervix during pregnancy is studied as an example to show how cervical thinning and dilation are generated by growth. An advanced hyperelasticity theory called morphoelasticity is employed to model the deformations, and a growth tensor is used to …

Fast Multipole Methods For Wave And Charge Source Interactions In Layered Media And Deep Neural Network Algorithms For High-Dimensional Pdes, Wenzhong Zhang

Mathematics Theses and Dissertations

In this dissertation, we develop fast algorithms for large scale numerical computations, including the fast multipole method (FMM) in layered media, and the forward-backward stochastic differential equation (FBSDE) based deep neural network (DNN) algorithms for high-dimensional parabolic partial differential equations (PDEs), addressing the issues of real-world challenging computational problems in various computation scenarios.

We develop the FMM in layered media, by first studying analytical and numerical properties of the Green's functions in layered media for the 2-D and 3-D Helmholtz equation, the linearized Poisson--Boltzmann equation, the Laplace's equation, and the tensor Green's functions for the time-harmonic Maxwell's equations and the …

A Fast Method For Computing Volume Potentials In The Galerkin Boundary Element Method In 3d Geometries, Sasan Mohyaddin

Mathematics Theses and Dissertations

We discuss how the Fast Multipole Method (FMM) applied to a boundary concentrated mesh can be used to evaluate volume potentials that arise in the boundary element method. If $h$ is the meshwidth near the boundary, then the algorithm can compute the potential in nearly $\Ord(h^{-2})$ operations while maintaining an $\Ord(h^p)$ convergence of the error. The effectiveness of the algorithms are demonstrated by solving boundary integral equations of the Poisson equation.

Finite Element Approximation Of Solutions Of The Equations Of Electroporoelasticity, Yu Hu

Mathematics Theses and Dissertations

In this thesis we consider the solution of the equations of electroporoelasticity, which are a combination of Maxwell's equations and the poroelasticity equations. Included is a description of suitable initial and boundary conditions, weak formulation of the equations, and the error estimate for a general numerical method.

Ensemble Data Fitting For Bathymetric Models Informed By Nominal Data, Samantha Zambo

Dissertations

Due to the difficulty and expense of collecting bathymetric data, modeling is the primary tool to produce detailed maps of the ocean floor. Current modeling practices typically utilize only one interpolator; the industry standard is splines-in-tension.

In this dissertation we introduce a new nominal-informed ensemble interpolator designed to improve modeling accuracy in regions of sparse data. The method is guided by a priori domain knowledge provided by artificially intelligent classifiers. We recast such geomorphological classifications, such as ‘seamount’ or ‘ridge’, as nominal data which we utilize as foundational shapes in an expanded ordinary least squares regression-based algorithm. To our knowledge …

Lower Bounds On Betti Numbers, Adam Boocher, Eloisa Grifo

Department of Mathematics: Faculty Publications

We survey recent results on bounds for Betti numbers of modules over polynomial rings, with an emphasis on lower bounds. Along the way, we give a gentle introduction to free resolutions and Betti numbers, and discuss some of the reasons why one would study these.

Computational Analysis To Study The Efficiency Of Shear Activated Nano-Therapeutics In The Treatment Of Atherosclerosis, Nicholas Jefopoulos

Theses, Dissertations and Culminating Projects

Strokes are the fifth leading cause of death in the United States and can cause long-term disabilities in patients who survive a stroke. The vast majority of these strokes are ischemic, primarily caused by intracranial atherosclerosis. Most therapies to combat intracranial atherosclerosis simply manage it and do not remove the buildup of plaque. Targeted shear-activated nanotherapeutics are currently being developed to remove these plaques. We discuss the roles that aggregate particle density, aggregate particle diameter, vessel geometry, stenosis shape and breakup threshold play in the efficiency of this new technology. Computational studies were performed to test these parameters in three …

Multilateration Index., Chip Lynch

Electronic Theses and Dissertations

We present an alternative method for pre-processing and storing point data, particularly for Geospatial points, by storing multilateration distances to fixed points rather than coordinates such as Latitude and Longitude. We explore the use of this data to improve query performance for some distance related queries such as nearest neighbor and query-within-radius (i.e. “find all points in a set P within distance d of query point q”). Further, we discuss the problem of “Network Adequacy” common to medical and communications businesses, to analyze questions such as “are at least 90% of patients living within 50 miles of a covered emergency …

The Food Truck Problem, Supply Chains And Extensions Of The Newsvendor Problem, Dennis Quayesam

Electronic Theses and Dissertations

Inventory control is important to ensuring sufficient quantities of items are available tomeet demands of customers. The Newsvendor problem is a model used in Operations Research to determine optimal inventory levels for fulfilling future demands. Our study extends the newsvendor problem to a food truck problem. We used simulation to show that the food truck does not reduce to a newsvendor problem if demand depends on exogenous factors such temperature, time etc. We formulate the food truck problem as a multi-product multi-period linear program and found the dual for a single item. We use Discrete Event Simulation to solve the …

Particle Trajectories In Shallow Water Models, Diana Torres

Theses and Dissertations

In this paper we will study particle trajectories under shallow water waves. We will examine equations such as the Korteweg-de Vries and systems dealing with Boussinesq and Euler's Equations to find relationships between particles irrotational velocities. Their solutions and behavior when modeling interacting surface waves will be explored. An attempt to find approximate solutions with different parameters, such as small amplitude and long-crested waves, that will lead to new information and study will be discussed.

Manifold Learning With Tensorial Network Laplacians, Scott Sanders

Electronic Theses and Dissertations

The interdisciplinary field of machine learning studies algorithms in which functionality is dependent on data sets. This data is often treated as a matrix, and a variety of mathematical methods have been developed to glean information from this data structure such as matrix decomposition. The Laplacian matrix, for example, is commonly used to reconstruct networks, and the eigenpairs of this matrix are used in matrix decomposition. Moreover, concepts such as SVD matrix factorization are closely connected to manifold learning, a subfield of machine learning that assumes the observed data lie on a low-dimensional manifold embedded in a higher-dimensional space. Since …

Applying Deep Learning To The Ice Cream Vendor Problem: An Extension Of The Newsvendor Problem, Gaffar Solihu

Electronic Theses and Dissertations

The Newsvendor problem is a classical supply chain problem used to develop strategies for inventory optimization. The goal of the newsvendor problem is to predict the optimal order quantity of a product to meet an uncertain demand in the future, given that the demand distribution itself is known. The Ice Cream Vendor Problem extends the classical newsvendor problem to an uncertain demand with unknown distribution, albeit a distribution that is known to depend on exogenous features. The goal is thus to estimate the order quantity that minimizes the total cost when demand does not follow any known statistical distribution. The …

A Deep Ensemble Method For Multi-Agent Reinforcement Learning: A Case Study On Air Traffic Control, Supriyo Ghosh, Sean Laguna, Laura Wynter, Laura Wynter, Hasan Poonawala

Research Collection School Of Computing and Information Systems

Reinforcement learning (RL), a promising framework for data-driven decision making in an uncertain environment, has successfully been applied in many real-world operation and control problems. However, the application of RL in a large-scale decentralized multi-agent environment remains a challenging problem due to the partial observability and limited communications between agents. In this paper, we develop a model-based kernel RL approach and a model-free deep RL approach for learning a decentralized, shared policy among homogeneous agents. By leveraging the strengths of both these methods, we further propose a novel deep ensemble multi-agent reinforcement learning (MARL) method that efficiently learns to arbitrate …

On A Stochastic Model Of Epidemics, Rachel Prather

Master's Theses

This thesis examines a stochastic model of epidemics initially proposed and studied by Norman T.J. Bailey [1]. We discuss some issues with Bailey's stochastic model and argue that it may not be a viable theoretical platform for a more general epidemic model. A possible alternative approach to the solution of Bailey's stochastic model and stochastic modeling is proposed as well. Regrettably, any further study on those proposals will have to be discussed elsewhere due to a time constraint.

A Modified Preconditioned Conjugate Gradient Method For Approximating The Scattering Amplitude, Samson Ayo

Master's Theses

In this thesis, we look at an iterative method for approximating the scattering amplitude that involves solving two linear systems: a forward system Ax=b and an adjoint system A^Ty=g. Once these two systems are solved, the scattering amplitude, defined by g^Tx=y^Tb is easily obtained.

We derive a conjugate gradient-like iteration for a nonsymmetric saddle point matrix that is constructed to have a real positive spectrum. We investigate the use of Schur Complement preconditioners with block-diagonal factorization to speed up the convergence of our method and compare …

Data Science And The Ice-Cream Vendor Problem, Makafui Azasoo

Electronic Theses and Dissertations

Newsvendor problems in Operations Research predict the optimal inventory levels necessary to meet uncertain demands. This thesis examines an extended version of a single period multi-product newsvendor problem known as the ice cream vendor problem. In the ice cream vendor problem, there are two products – ice cream and hot chocolate – which may be substituted for one another if the outside temperature is no too hot or not too cold. In particular, the ice cream vendor problem is a data-driven extension of the conventional newsvendor problem which does not require the assumption of a specific demand distribution, thus allowing …