Multi-scale parameter estimation for the steady state diffusion equation /

Parameter estimation problems where the unknown quantities

Bibliographic Details
Main Author: Zheng, Yan
Format: Thesis Book
Language:English
Published: [Place of publication not identified] : [publisher not identified] ; 1997.
Subjects:
Online Access:http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=736584771&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD
Description
Summary:Parameter estimation problems where the unknown quantities
have both discontinuous and spatially heterogeneous behavior
on multiple length scales are very challenging,
mathematically and computationally. For example, in flow
through layered (or fractured) porous media, both porosity
and absolute permeability will have spatial variability (and
discontinuties) at multiple length scales. One can model
this by representing the porosity and absolute permeability
by a set of basis functions which incorporate scale
dependence. One example of such a basis are the Daubechies
wavelets of order N. The simplest such basis (Harr) is used
for recovering discontinuous coefficients. By comparing the
output of the numerical simulation and experimental data
from the physical system itself, constraints on the unknowns
results. When the unknowns are present at several length
scales, standard least squares methods converge slowly, if
at all. By representing coefficients hierarchically, and
distributing the degree of freedom in an intelligent
fashion, the least squares method become much more robust.
A complete convergence analysis and error estimation are
given. Extensive numerical experiments are implemented.
The numerical results validate the convergence analysis.
Item Description:Vita.
"Major Subject: Mathematics".
Physical Description:viii, 101 leaves : illustrations ; 28 cm.
Issued also on microfiche from University Microfilms Inc.
Bibliography:Includes bibliographical references: pages 89-96.