Convergence Rates of Spectral Distribution of Random Inner Product Kernel Matrices
AdvisorRider, Brian (Brian C.)
Committee memberYang, Wei-shih, 1954-
Random Geometric Graph
Random Inner Product Kernel Matrix
Permanent link to this recordhttp://hdl.handle.net/20.500.12613/3132
MetadataShow full item record
AbstractThis dissertation has two parts. In the first part, we focus on random inner product kernel matrices. Under various assumptions, many authors have proved that the limiting empirical spectral distribution (ESD) of such matrices A converges to the Marchenko- Pastur distribution. Here, we establish the corresponding rate of convergence. The strategy is as follows. First, we show that for z = u + iv ∈ C, v > 0, the distance between the Stieltjes transform m_A (z) of ESD of matrix A and Machenko-Pastur distribution m(z) is of order O (log n \ nv). Next, we prove the Kolmogorov distance between ESD of matrix A and Marchenko-Pastur distribution is of order O(3\log n\n). It is the less sharp rate for much more general class of matrices. This uses a Berry-Esseen type bound that has been employed for similar purposes for other families of random matrices. In the second part, random geometric graphs on the unit sphere are considered. Observing that adjacency matrices of these graphs can be thought of as random inner product matrices, we are able to use an idea of Cheng-Singer to establish the limiting for the ESD of these adjacency matrices.
ADA complianceFor Americans with Disabilities Act (ADA) accommodation, including help with reading this content, please contact email@example.com
Showing items related by title, author, creator and subject.
Going beyond the Random Phase Approximation: A systematic assessment of structural phase transitions and interlayer binding energiesRuzsinszky, Adrienn; Perdew, John P.; Yan, Qimin; Matsika, Spiridoula (Temple University. Libraries, 2018)The Random Phase Approximation and beyond Random Phase Approximation methods based on Adiabatic Connection Fluctuation Dissipation Theorem (ACFD) are tested for structural phase transitions of different groups of materials, including metal to metal, metal to semiconductor, semiconductor to semiconductor transitions. Also the performance assessment of semilocal density functionals with or without empirical long range dispersion corrections has been explored for the same cases. We have investigated the structural phase transitions of three broad group of materials, semi- conductor to metal transitions involving two symmetric structures, semiconductor to metal and wide bandgap semiconductor to semiconductor transitions involving at least one lower symmetric structure and lastly special cases comprising metal to metal transitions and transitions between energetically very close structural phases. The first group contains Si (diamond → β-tin), Ge (diamond → β-tin) and SiC (zinc blende → rocksalt), second group contains GaAs (zinc blende → cmcm) and SiO 2 (quartz → stishovite) and third group contains Pb (fcc → hcp), C(graphite → diamond) and BN (cubic → hexagonal) respectively. We have found that the difference in behavior of exchange and correlation in semilocal functionals and ACFD methods is striking. For the former, the exchange potential and energy often comprise the majority of the binding described by density functional approximations, and the addition of the correlation energy and potential often induce only a (relatively) small shift from the exchange- only results. For the ACFD, however, non self-consistent EXX typically underbinds by a considerable degree resulting in wildly inaccurate results. Thus the addition of correlation leads to very large shifts in the exchange-only results, in direct contrast to semilocal correlation. This difference in behavior is directly linked to the non-local nature of the EXX, and even though the exchange-only starting point is often nowhere close to experiment, the non-local correlation from the ACFD corrects this deficiency and yields the missing binding needed to produce accurate results. Thus we find the ACFD approach to be vital in the validation of semilocal results and recommend its use in materials where experimental results cannot be straightforwardly compared to other approximate electronic structure calculations. Utilizing the second-order approximation to Random Phase Approximation renormalized (RPAr) many-body perturbation theory for the interacting density-density response function, we have used a so-called higher-order terms (HOT) approximation for the correlation energy. In combination with the first-order RPAr correction, the HOT method faithfully captures the infinite- order correlation for a given exchange-correlation kernel, yielding errors of the total correlation energy on the order of 1% or less for most systems. For exchange-like kernels, our new method has the further benefit that the coupling-strength integration can be completely eliminated resulting in a modest reduction in computational cost compared to the traditional approach. When the correlation energy is accurately reproduced by the HOT approximation, structural properties and energy differences are also accurately reproduced, as confirmed by finding interlayer binding energies of several periodic solids and compared that to some molecular systems along with some phase transition parameters of SiC. Energy differences involving fragmentation have proved to be challenging for the HOT method, however, due to errors that do not cancel between a composite system and its constituent pieces which has been verified in our work as well.
A Nonparametric Test for Deviation from RandomnessIglewicz, Boris; Chitturi, Pallavi; Dong, Yuexiao; Chervoneva, Inna (Temple University. Libraries, 2012)There are many existing tests used to determine if a series consists of a random sample. Often these tests have restrictive distributional assumptions, size distortions, or low power for key useful alternative situations. The interest of this dissertation lies in developing an alternative nonparametric test to determine whether a series consists of a random sample. The proposed test detects deviations from randomness, without a priori distributional assumption, when observations are not independent and identically distributed (i.i.d.), which is suitable for our motivating stock market index data. Departures from i.i.d. are tested by subdividing data into subintervals and then using a conditional probability measure within intervals as a binomial test. This nonparametric test is designed to detect deviations of neighboring observations from randomness when the data set consists of time series observations. Simulation results confirm correct test size for varied distributions and good power for detecting alternative cases. This test is compared to a number of other popular methods and shown to be a competitive alternative. Although the proposed test may be applicable to multiple areas, this dissertation is mostly interested in applications to stock market and regression data. The proposed test is effectively illustrated with the common three stock market index data sets using a newly created transformation, and shown to perform exceptionally well.
Effects of a Breakfast in the Classroom Initiative on Childhood Overweight and Obesity in Low-Income, Ethnically Diverse Youth: A Randomized Control TrialDavey, Adam; Fisher, Jennifer O. (Temple University. Libraries, 2017)Background: School districts across the country have adopted breakfast in the classroom (BIC) initiatives as a means of increasing participation in the School Breakfast Program (SBP). Little is known regarding the impact of such programs on children’s weight status. This study sought to evaluate the impact of a BIC initiative on the combined incidence of overweight/obesity among urban school aged children. Design: Cluster-randomized controlled trial with matched school pairs. Intervention schools received the “One Healthy Breakfast” program included BIC, nutrition-education, social marketing, corner store marketing, and parent outreach, while control schools continued to serve breakfast in the cafeteria Baseline data were collected in October-December, 2013, with midpoint and endpoint data collected in May-June 2015 and May-June 2016 respectively. Schools were matched based on school size, food service type, and racial/ethnic composition. Setting/Participants: 1371 4th-6th grade studen