Dr. Martin Gutting's Personal Homepage

 

Inverse Problems

Lecture (4 SWS): every Monday  10:00 - 11:30 in 44-465 and every Wednesday  8:15 - 9:45 in 48-538 (as noted in KIS), first lecture on 20.04.2009
The tutorials take place on every Wednesday 13:45 - 15:15 in 49-506.
Lecturer: Martin Gutting
The lecture will be given in English!

News

Tutorials 10 and 11 are shifted to Thursday, July 16 and 23, 11:45-13:15 in 49-506.
Illustrations corresponding to regularization by TSVD and Tikhonov-Phillips  (only as PDF)
Illustrations corresponding to the heat equation
Illustrations corresponding to differentiation
Illustrations corresponding to reconstructions via singular value decomposition with noise
Illustrations of spherical harmonics
Fundamentals in Functional Analysis: download as PDF or as Postscript (ca. 0.6 MB)

Lecture Notes as PDF: lecture notes (July 22) Note: the file is currently roughly 9 MB.

Exercises for the tutorials

sheet 1
sheet 2 Solution for Exercise 2.4 (.m-file)
sheet 3
sheet 4
sheet 5
sheet 6
sheet 7 Solution for Exercise 7.2 (PDF with all the figures) (zip-file containing the m-files)
sheet 8
sheet 9
sheet 10
sheet 11

Contents

Inverse problems today appear in many technological problems. If one wants to know the reason of a measured effect, one has to cope with an inverse problem. For example in computer tomography, the attenuation of x-rays is measured after they have passed the object of interest (e.g. the human body). The cause for the attenuation is the density of the object. From the mathematical point of view, inverse problems consist of inverting certain operator equations of the first kind. In the equation Ax=y, y is given and x is wanted. An inverse problem is especially concerned with the case that y is not given exactly, y is not in the image of A or A^-1 is not continuous, which is the case if A is compact and D(A) is not finite dimensional. Inverse problems have to be regularized (stabilized), to avoid the appearing amplification of errors. The lecture will give a mathematical introduction for the solution and the regularization of inverse problems with concrete geomathematical and other technological applications.

Inhalt

Inverse Probleme treten in der heutigen Technologie häufig auf. Immer wenn man von einer beobachteten Wirkung auf deren Ursache schließen möchte, liegt ein inverses Problem vor. So wird in der Computer-Tomographie die Abminderung von Röntgenstrahlen gemessen beim Durchgang durch ein Objekt (z.B. menschlicher Körper). Die Ursache der Abminderung ist die Dichte des Objekts. Aus mathematischer Sicht bestehen inverse Probleme darin, Operatorgleichungen der ersten Art zu lösen. Man habe eine Gleichung Ax=y und y sei gegeben. Gesucht ist x. Hierbei ergibt sich ein inverses Probleme u. a. dadurch, dass y nur ungenau angegeben ist oder dass y nicht im Bildraum von A liegt oder dass A^-1 zwar existiert, aber nicht stetig ist, was der Fall ist, wenn A kompakt ist und D(A) nicht endlich dimensional ist. Inverse Probleme müssen regularisiert (stabilisiert) werden, um die auftretende Fehlerverstärkung zu vermeiden. Die Vorlesung führt ein in die mathematischen Grundlagen zur Lösung und Regularisierung inverser Probleme, zielt dabei aber auch auf konkrete Anwendungen aus der Geomathematik und anderer technischer Gebiete ab.

Overview

  1. What are inverse problems? Examples, abstract formulation, characteristic behavior.
  2. Fundamentals of Functional Analysis.
  3. Operator Equations: Moore-Penrose Inverse, compact operators, singular value decomposition, examples: SST and SGG.
  4. Regularization of Linear Problems: regularizing filters, classification of regularization methods, truncated singular value decomposition, Tikhonov-Philipps, strategies for the choice of the regularization parameter.
  5. Generalized Tikhonov-Philipps regularization, total variation regularization, iterated Tikhonov-Philipps regularization.
  6. Multiscale regularization, regularizing scaling functions and wavelets.
  7. Iterative regularization methods: Landweber iteration, semi-iterative methods, conjugate grgadient method
  8. Projection Methods and Discretization

Literature

  1. Louis, A.: Inverse und schlecht gestellte Probleme. Teubner 1989.
  2. Colton, D., Kress, R.: Inverse Acoustic and Electromagnetic Scattering Theory. Springer 1992.
  3. Kirsch, A.: An Introduction to the Mathematical Theory of Inverse Problems. Springer 1996.
  4. Rieder, A.: Keine Probleme mit inversen Problemen. Vieweg 2003.
Further references are given in the lecture notes.