Back to search

National and Kapodistrian University of Athens

Μεθοδολογίες υπολογισμού διγραμμικών μορφών και εφαρμογές

Abstract

dc:description

The management of large-scale data is necessary nowadays and the development of appropriate methodologies is required. The computation of bilinear forms and related quantities leads to useful results and it is often a criterion for properties related to the corresponding application. Most of the scientific applications such as network analysis, computer science, statistics, economics, etc., require the management of large-scale matrices for the extraction of useful results. In these applications, it is usually not possible the direct computation of the desired bilinear forms and thus it is necessary their estimation.In the present PhD thesis, we derive estimates for the bilinear forms of the type xᵀf(A)y, where A ϵ Rᵖˣᵖ is a given matrix with specific structure, x, y ϵ Rᵖ and f is a smooth function defined on the spectrum of the matrix A. The developed method is based on an extrapolation procedure. Specifically, we generate estimates for the bilinear form xᵀA⁻¹y, where the matrix A is invertible. If the vectors x, y are chosen to be columns of the identity matrix, we can obtain estimates for the elements of the inverse of the given matrix A. We also adjust some numerical methods based on an extrapolation procedure and the Gaussian quadrature rules to estimate the diagonal of the function of symmetric matrices, without approximating the whole matrix f(A).Moreover, we derive families of one, two, and three term vector estimates for the action of the matrix f(A) on a given vector b, i.e. the product f(A)b. The calculation of the quantity f(A)b arises from applications in which it is not necessary to estimate the whole matrix f(A) or it is not feasible to calculate this matrix, especially when A is large. We also deal with the estimation and minimization of the generalized cross-validation function (GCV function) by combining an extrapolation procedure and a statistical approach. The GCV function is a popular tool for determining the tuning parameter in the linear regression model or equivalently the regularization parameter in the Tikhonov regularization. Finally, we apply the generated estimates for the aforementioned quantities in problems which deal with network analysis. Specifically, we will determine the most important nodes of a network and the ease of traveling between nodes. We also use the derived estimates for the GCV function for solving discrete ill-posed problems as well as for estimating the coefficients of the linear regression model which is used in Statistics.

Degree

thesis:*
Grantor dc:publisher
National and Kapodistrian University of Athens
Year dc:date
2020

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Roupa, Paraskevi
  • Ρούπα, Παρασκευή

Subjects

dc:subject × 13

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/47766

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Roupa, Paraskevi; Ρούπα, Παρασκευή. Μεθοδολογίες υπολογισμού διγραμμικών μορφών και εφαρμογές. National and Kapodistrian University of Athens, 2020. http://hdl.handle.net/10442/hedi/47766