Skip to main content

1994 | Buch

Ill-Posed Problems: Theory and Applications

verfasst von: A. Bakushinsky, A. Goncharsky

Verlag: Springer Netherlands

Buchreihe : Mathematics and Its Applications

insite
SUCHEN

Über dieses Buch

Recent years have been characterized by the increasing amountofpublications in the field ofso-called ill-posed problems. This is easilyunderstandable because we observe the rapid progress of a relatively young branch ofmathematics, ofwhich the first results date back to about 30 years ago. By now, impressive results have been achieved both in the theory ofsolving ill-posed problems and in the applicationsofalgorithms using modem computers. To mention just one field, one can name the computer tomography which could not possibly have been developed without modem tools for solving ill-posed problems. When writing this book, the authors tried to define the place and role of ill­ posed problems in modem mathematics. In a few words, we define the theory of ill-posed problems as the theory of approximating functions with approximately given arguments in functional spaces. The difference between well-posed and ill­ posed problems is concerned with the fact that the latter are associated with discontinuous functions. This approach is followed by the authors throughout the whole book. We hope that the theoretical results will be of interest to researchers working in approximation theory and functional analysis. As for particular algorithms for solving ill-posed problems, the authors paid general attention to the principles ofconstructing such algorithms as the methods for approximating discontinuous functions with approximately specified arguments. In this way it proved possible to define the limits of applicability of regularization techniques.

Inhaltsverzeichnis

Frontmatter
1. General problems of regularizability
Abstract
Any mathematical model sets some correspondence between two kinds of objects. One class of objects includes characteristics of the model, and the other consists of experimentally observed attributes of the studied phenomena. The problems of processing experimental data are always concerned with inevitable experimental errors. For the purposes of present analysis, we shall consider the objects of the second kind belonging to some metric space X, and of the first kind - to space Y. The mathematical model establishes the certain correspondence y =G(x) between input dataxXand characteristics of the model yY. Modeling is aimed at obtaining model characteristics y using approximately specified input datax.
A. Bakushinsky, A. Goncharsky
Chapter 2. Regularizing algorithms on compacta
Abstract
Henceforth we consider how to construct regularizing algorithms for solving the ill-posed problems. The typical ill-posed problem is the operator equation of the first kind
$$ Az = u,z \in U $$
(2.1)
.
A. Bakushinsky, A. Goncharsky
3. Tikhonov’s scheme for constructing regularizing algorithms
Abstract
In this chapter the most popular scheme for constructing regularizing algorithms is discussed. In fact, this scheme has started the modern stage of development of the numerical methods for solving the ill-posed problems. The constructions described in this chapter immediatly lead to easily implementable numerical algorithms widely used in practice for the linear problems. For the non-linear problems, Tikhonov’s scheme may be used as a principal base for creating efficient numerical algorithms.
A. Bakushinsky, A. Goncharsky
4. General technique for constructing linear RA for linear problems in Hilbert space
Abstract
Speaking of linear ill-posed problems in Hilbert spaces, we henceforth bear in mind the two following particular problems: Problem 1. Solve the operator equation
$$ Az = u,z \in Z,u \in U $$
(4.1)
with a closed and, generally speaking, uninvertible operator A. The domain of definition of A DA is dense inZ.
A. Bakushinsky, A. Goncharsky
5. Iterative algorithms for solving non-linear ill-posed problems with monotonic operators. Principle of iterative regularization
Abstract
In this chapter we discuss schemes of constructing iterative approximations for the wide range of non-linear mappings originating from the non-linear ill-posed problems, as well as the regularizing algorithms based on such approximations. The traditional variational technique of constructing approximations (see Chapter 3) though being applicable to non-linear case, does not directly lead to numerically implementable RA. This occurs due to additional approximations (usually the iterative ones) which are necessary to calculate values of operatorsR a (orRδ).
A. Bakushinsky, A. Goncharsky
6. Applications of the principle of iterative regularization
Abstract
In this chapter we discuss the opportunities unveiled by the iterative regularization technique in the traditional applications of computer science. This technique allows to construct theoretically approved iterative algorithms in circumstances when the classical approaches are either inapplicable, or associated with considerable difficulties.
A. Bakushinsky, A. Goncharsky
7. Iterative methods for solving non-linear ill-posed operator equations with non-monotonic operators
Abstract
For numerous applied problems the condition of operator being monotonic (which is assumed everywhere in Chapters 5 and 6) proves to be too restrictive. The analysis given in this chapter does not use the assumption of monotony. As usual, we concentrate on the cases when the conditions ensuring convergence of traditional iterative techniques are not satisfied.
A. Bakushinsky, A. Goncharsky
8. Application of regularizing algorithms to solving practical problems
Abstract
In the previous chapters of this book we have advanced and mathematically justified various techniques designed to solve the wide range of the ill-posed problems. Numerical implementation of these techniques can be successfully applied to mathematical modelling as well as to solving the inverse problems. In this chapter we present only a few examples characterizing the opportunities of the mathematical tools described in this book.
A. Bakushinsky, A. Goncharsky
Backmatter
Metadaten
Titel
Ill-Posed Problems: Theory and Applications
verfasst von
A. Bakushinsky
A. Goncharsky
Copyright-Jahr
1994
Verlag
Springer Netherlands
Electronic ISBN
978-94-011-1026-6
Print ISBN
978-94-010-4447-9
DOI
https://doi.org/10.1007/978-94-011-1026-6