On inverse problems with partial data.
Salle Euler Violet – Tuesday, June 14, 2016 at 02:00 PM.
Many physical problems modelled by differential or integral equations require data which are either not completely available from measurements or incompatible with the model due to noise. This situation raises an issue of efficient interpolation/extrapolation and approximation. My thesis includes investigation of 3 different problems that fit in this framework: recovery of harmonic functions from partial boundary data, construction of efficient bases which are eigenfunctions of the truncated Poisson integral operator, and, on more practical side, derivation of the formulas for the net magnetization of a sample in terms of the partial measurements of the field.
Eduardo Andrade Lima, Massachusetts Institute of Technology (MIT), Boston, USA (examinateur)
Didier Clamond, University of Nice – Sophia Antipolis (examinateur)
Nuutti Hyvonen, Aalto University, Finlande (rapporteur)
Emmanuel RUSS, University Grenoble Alpes, Institut Fourier (rapporteur)
Laurent Baratchart, INRIA, équipe Apics (directeur de thèse)
Juliette Leblond, INRIA, équipe Apics (directrice de thèse)