Organizational Unit:
School of Mathematics
School of Mathematics
Permanent Link
Research Organization Registry ID
Description
Previous Names
Parent Organization
Parent Organization
Organizational Unit
Includes Organization(s)
ArchiveSpace Name Record
Publication Search Results
Now showing
1 - 10 of 120
-
ItemNumerical algorithms based on the back and forth error compensation and correction(Georgia Institute of Technology, 2014-12-18) Hu, LiliIn this thesis we carry out a further study of the back and forth error compensation and correction (BFECC) method. The first part discusses the time reversibility of numerical schemes. Motivated by the BFECC method, a variety of new numeri- cal schemes that aim at improving the time reversibility are developed and studied. We also introduce an interpolation algorithm based on BFECC in this part. In the second part we introduce a new limiting strategy which requires another backward advection in time so that overshoots/undershoots at the new time level get exposed when they are transformed back to compare with the solution at the old time level. This new technique is very simple to implement even for unstructured meshes and is able to eliminate artifacts induced by jump discontinuities in the solution itself or in its derivatives.
-
ItemSome results on sums and products(Georgia Institute of Technology, 2014-11-17) Pryby, Christopher IanWe demonstrate new results in additive combinatorics, including a proof of a conjecture by J. Solymosi: for every epsilon > 0, there exists delta > 0 such that, given n² points in a grid formation in R², if L is a set of lines in general position such that each line intersects at least n^{1-delta} points of the grid, then |L| < n^epsilon. This result implies a conjecture of Gy. Elekes regarding a uniform statistical version of Freiman's theorem for linear functions with small image sets.
-
ItemAccelerated algorithms for composite saddle-point problems and applications(Georgia Institute of Technology, 2014-11-13) He, YunlongThis dissertation considers the composite saddle-point (CSP) problem which is motivated by real-world applications in the areas of machine learning and image processing. Two new accelerated algorithms for solving composite saddle-point problems are introduced. Due to the two-block structure of the CSP problem, it can be solved by any algorithm belonging to the block-decomposition hybrid proximal extragradient (BD-HPE) framework. The framework consists of a family of inexact proximal point methods for solving a general two-block structured monotone inclusion problem which, at every iteration, solves two prox sub-inclusions according to a certain relative error criterion. By exploiting the fact that the two prox sub-inclusions in the context of the CSP problem are equivalent to two composite convex programs, the first part of this dissertation proposes a new instance of the BD-HPE framework that approximately solves them using an accelerated gradient method. It is shown that this new instance has better iteration-complexity than the previous ones. The second part of this dissertation introduces a new algorithm for solving a special class of CSP problems. The new algorithm is a special instance of the hybrid proximal extragradient (HPE) framework in which a Nesterov's accelerated variant is used to approximately solve the prox subproblems. One of the advantages of the this method is that it works for any constant choice of proximal stepsize. Moreover, a suitable choice of the latter stepsize yields a method with the best known (accelerated inner) iteration complexity for the aforementioned class of saddle-point problems. Experiment results on both synthetic CSP problems and real-world problems show that the two method significantly outperform several state-of-the-art algorithms.
-
ItemMultiscale and stochastic methods for inverse source problems and signal analysis(Georgia Institute of Technology, 2014-09-30) Zhou, Haomin
-
ItemHow Quantum Theory and Statistical Mechanics Gave a Polynomial of Knots(Georgia Institute of Technology, 2014-09-25) Jones, VaughanWe will see how a result in von Neumann algebras (a theory developed by von Neumann to give the mathematical framework for quantum physics) gave rise, rather serendipitously, to an elementary but very useful invariant in the theory of ordinary knots in three dimensional space. Then we'll look at some subsequent developments of the theory, and talk about a thorny problem which remains open.
-
ItemA non-asymptotic study of low-rank estimation of smooth kernels on graphs(Georgia Institute of Technology, 2014-07-23) Rangel Walteros, Pedro AndresThis dissertation investigates the problem of estimating a kernel over a large graph based on a sample of noisy observations of linear measurements of the kernel. We are interested in solving this estimation problem in the case when the sample size is much smaller than the ambient dimension of the kernel. As is typical in high-dimensional statistics, we are able to design a suitable estimator based on a small number of samples only when the target kernel belongs to a subset of restricted complexity. In our study, we restrict the complexity by considering scenarios where the target kernel is both low-rank and smooth over a graph. Using standard tools of non-parametric estimation, we derive a minimax lower bound on the least squares error in terms of the rank and the degree of smoothness of the target kernel. To prove the optimality of our lower-bound, we proceed to develop upper bounds on the error for a least-square estimator based on a non-convex penalty. The proof of these upper bounds depends on bounds for estimators over uniformly bounded function classes in terms of Rademacher complexities. We also propose a computationally tractable estimator based on least-squares with convex penalty. We derive an upper bound for the computationally tractable estimator in terms of a coherence function introduced in this work. Finally, we present some scenarios wherein this upper bound achieves a near-optimal rate. The motivations for studying such problems come from various real-world applications like recommender systems and social network analysis.
-
ItemLinear systems on metric graphs and some applications to tropical geometry and non-archimedean geometry(Georgia Institute of Technology, 2014-07-02) Luo, YeThe divisor theories on finite graphs and metric graphs were introduced systematically as analogues to the divisor theory on algebraic curves, and all these theories are deeply connected to each other via tropical geometry and non-archimedean geometry. In particular, rational functions, divisors and linear systems on algebraic curves can be specialized to those on finite graphs and metric graphs. Important results and interesting problems, including a graph-theoretic Riemann-Roch theorem, tropical proofs of conventional Brill-Noether theorem and Gieseker-Petri theorem, limit linear series on metrized complexes, and relations among moduli spaces of algebraic curves, non-archimedean analytic curves, and metric graphs are discovered or under intense investigations. The content in this thesis is divided into three main subjects, all of which are based on my research and are essentially related to the divisor theory of linear systems on metric graphs and its application to tropical geometry and non-archimedean geometry. Chapter 1 gives an overview of the background and a general introduction of the main results. Chapter 2 is on the theory of rank-determining sets, which are subsets of a metric graph that can be used for the computation of the rank function. A general criterion is provided for rank-determining sets and certain specific examples of finite rank-determining sets are presented. Chapter 3 is on the subject of a tropical convexity theory on linear systems on metric graphs. In particular, the notion of general reduced divisors is introduced as the main tool used to study this tropical convexity theory. Chapter 4 is on the subject of smoothing of limit linear series of rank one on re_ned metrized complexes. A general criterion for smoothable limit linear series of rank 1 is presented and the relations between limit linear series of rank 1 and possible harmonic morphisms to genus 0 metrized complexes are studied.
-
ItemA numerical study of vorticity-enhanced heat transfer(Georgia Institute of Technology, 2014-06-26) Wang, XiaolinIn this work, we have numerically studied the effect of the vorticity on the enhancement of heat transfer in a channel flow. In the first part of the work, we focus on the investigation of a channel flow with a vortex street as the incoming flow. We propose a model to simulate the fluid dynamics. We find that the flow exhibits different properties depending on the value of four dimensionless parameters. In particularly, we can classify the flows into two types, active and passive vibration, based on the sign of the incoming vortices. In the second part of the work, we discuss the heat transfer process due to the flows just described and investigate how the vorticity in the flow improves the efficiency of the heat transfer. The temperature shows different characteristics corresponding to the active and passive vibration cases. In active vibration cases, the vortex blob improves the heat transfer by disrupting the thermal boundary layer and preventing the decay of the wall temperature gradient throughout the channel, and by enhancing the forced convection to cool down the wall temperature. The heat transxfer performance is directly related to the strength of the vortex blobs and the background flow. In passive vibration cases, the corresponding heat transfer process is complicated and varies dramatically as the flow changes its properties. We also studied the effect of thermal parameters on heat transfer performance. Finally, we propose a more realistic optimization problem which is to minimize the maximum temperature of the solids with a given input energy. We find that the best heat transfer performance is obtained in the active vibration case with zero background flow.
-
ItemStein fillings of contact structures supported by planar open books(Georgia Institute of Technology, 2014-06-20) Kaloti, AmeyIn this thesis we study topology of symplectic fillings of contact manifolds supported by planar open books. We obtain results regarding geography of the symplectic fillings of these contact manifolds. Specifically, we prove that if a contact manifold (M,ξ) is supported by a planar open book, then Euler characteristic and signature of any Stein filling of (M,ξ) is bounded. We also prove a similar finiteness result for contact manifolds supported by spinal open books with planar pages. Moving beyond the geography of Stein fillings, we classify fillings of some lens spaces. In addition, we classify Stein fillings of an infinite family of contact 3-manifolds up to diffeomorphism. Some contact 3-manifolds in this family can be obtained by Legendrian surgeries on (S³, ξ std) along certain Legendrian 2-bridge knots. We also classify Stein fillings, up to symplectic deformation, of an infinite family of contact 3-manifolds which can be obtained by Legendrian surgeries on (S³, ξ std) along certain Legendrian twist knots. As a corollary, we obtain a classification of Stein fillings of an infinite family of contact hyperbolic 3-manifolds up to symplectic deformation.
-
ItemInvariant densities for dynamical systems with random switching(Georgia Institute of Technology, 2014-06-19) Hurth, TobiasWe studied invariant measures and invariant densities for dynamical systems with random switching (switching systems, in short). These switching systems can be described by a two-component Markov process whose first component is a stochastic process on a finite-dimensional smooth manifold and whose second component is a stochastic process on a finite collection of smooth vector fields that are defined on the manifold. We identified sufficient conditions for uniqueness and absolute continuity of the invariant measure associated to this Markov process. These conditions consist of a Hoermander-type hypoellipticity condition and a recurrence condition. In the case where the manifold is the real line or a subset of the real line, we studied regularity properties of the invariant densities of absolutely continuous invariant measures. We showed that invariant densities are smooth away from critical points of the vector fields. Assuming in addition that the vector fields are analytic, we derived the asymptotically dominant term for invariant densities at critical points.