In this note, we consider an inverse problem in a jump diffusion model. Using characteristic functions we prove the injectivity of the forward operator mapping the five parameters determining the model to the density function of the return distribution.

Interest in elastographic techniques for soft tissue imaging has grown as relevant research continues to indicate a correlation between tissue histology and mechanical stiffness. Digital Image Elasto-Tomography (DIET) presents a novel method for identifying cancerous lesions via a three-dimensional image of elastic properties. Stiffness reconstruction with DIET takes steady-state motion captured with a digital camera array as the input to an elastic property reconstruction algorithm, where finite element methods allow simulation of phantom motion at a range of internal stiffness distributions. The low cost and high image contrast achievable with a DIET system may be particularly suited to breast cancer screening, where traditional modalities such as mammography have issues with limited sensitivity and patient discomfort. Proof of concept studies performed on simulated data sets confirmed the potential of the DIET technique, leading to the development of an experimental apparatus for surface motion capture from a range of soft tissue approximating phantoms. Error studies performed on experimental data from these phantoms using a limited number of shape and modulus parameters indicated that accurate measurements of surface motion provide sufficient information to identify a stiffness distribution in both homogeneous and heterogeneous cases. The elastic reconstruction performed on simulated and experimental data considered both deterministic and stochastic algorithms, with a combination of the two approaches found to give the most accurate results, for a realistic increase in computational cost. The reconstruction algorithm developed has the ability to successfully resolve a hard spherical inclusion within a soft phantom, and in addition demonstrated promise in reconstructing the correct stiffness distribution when no inclusion is present.

Reconstruction of continuous signals from a number of their discrete samples is central to digital signal processing. Digital devices can only process discrete data and thus processing the continuous signals requires discretization. After discretization, possibility of unique reconstruction of the source signals from their samples is crucial. The classical sampling theory provides bounds on the sampling rate for unique source reconstruction, known as the Nyquist sampling rate. Recently a new sampling scheme, Compressive Sensing (CS), has been formulated for sparse signals. CS is an active area of research in signal processing. It has revolutionized the classical sampling theorems and has provided a new scheme to sample and reconstruct sparse signals uniquely, below Nyquist sampling rates. A signal is called (approximately) sparse when a relatively large number of its elements are (approximately) equal to zero. For the class of sparse signals, sparsity can be viewed as prior information about the source signal. CS has found numerous applications and has improved some image acquisition devices. Interesting instances of CS can happen, when apart from sparsity, side information is available about the source signals. The side information can be about the source structure, distribution, etc. Such cases can be viewed as extensions of the classical CS. In such cases we are interested in incorporating the side information to either improve the quality of the source reconstruction or decrease the number of the required samples for accurate reconstruction. A general CS problem can be transformed to an equivalent optimization problem. In this thesis, a special case of CS with side information about the feasible region of the equivalent optimization problem is studied. It is shown that in such cases uniqueness and stability of the equivalent optimization problem still holds. Then, an efficient reconstruction method is proposed. To demonstrate the practical value of the proposed scheme, the algorithm is applied on two real world applications: image deblurring in optical imaging and surface reconstruction in the gradient field. Experimental results are provided to further investigate and confirm the effectiveness and usefulness of the proposed scheme.

In this thesis, we study inverse problems of restoration of the unknown function in a boundary condition, where on the boundary of the domain there is a convective heat exchange with the environment. Besides the temperature of the domain, we seek either the temperature of the environment in Problem I and II, or the coefficient of external boundary heat emission in Problem III and IV. An additional information is given, which is the overdetermination condition, either on the boundary of the domain (in Problem III and IV) or on a time interval (in Problem I and II). If solution of inverse problem exists, then the temperature can be defined everywhere on the domain at all instants. The thesis consists of six chapters. In the first chapter, there is the introduction where the definition and applications of inverse problems are given and definition of the four inverse problems, that we will analyze in this thesis, are stated. In the second chapter, some definitions and theorems which we will use to obtain some conclusions about the corresponding direct problem of our four inverse problems are stated, and the conclusions about direct problem are obtained. In the third, fourth, fifth and sixth chapters we have the analysis of inverse problems I, II, III and IV, respectively.

Cette thèse porte sur la construction de modèles stochastiques, régionaux et globaux du champ magnétique sur les quatre derniers millénaires à l'aide de mesures archéomagnétiques. Ces données présentent une répartition spatiale et temporelle très inhomogène, et sont caractérisées par de fortes incertitudes sur la mesure et sur la date. La reconstruction du champ constitue alors un problème inverse mal posé. Afin de déterminer la solution la plus adaptée, une information a priori sur le modèle doit être choisie. Elle consiste généralement en une régularisation arbitraire du champ magnétique (lissage en temps et en espace). Contrairement aux études précédentes, nous utilisons les statistiques temporelles du champ magnétique, dérivées des données d'observatoires, satellitaires et paléomagnétiques pour définir l'information a priori via des fonctions d'auto-covariances. Ces statistiques sont confirmées par des résultats issus de simulations numériques. Cette méthode bayésienne permet de s'affranchir de l'utilisation de fonctions supports arbitraires comme les splines pour l'interpolation temporelle. Le résultat final consiste en un ensemble de réalisations possibles du champ magnétique dont la dispersion caractérise l'incertitude sur le modèle. Afin de prendre en compte les erreurs de datation, nous développons par ailleurs une méthode basée sur l'utilisation de Markov Chain Monte Carlo (MCMC). Elle nous permet d'explorer de manière efficace l'espace des dates possibles et ainsi de sélectionner les modèles les plus probables. Cette méthode est une amélioration de la méthode de bootstrap classique, qui donne le même poids à des tirages aléatoires de dates présentant des probabilités très variables. Les ensembles de réalisations sélectionnés par la méthode MCMC aboutissent à la construction d'une densité de probabilités en lieu et place d'une courbe unique. La méthode bayésienne combinée à la méthode Markov Chain Monte Carlo nous a permis de construire des courbes régionales présentant des variations plus rapides que celles obtenues par d'autres études. Les courbes représentées sous forme de densités de probabilités ne sont pas nécessairement gaussiennes, et la méthode permet d'affiner l'estimation de l'âge de chacune des observations. La méthode bayésienne a été utilisée pour la construction de modèles globaux pour lesquels le dipôle axial présente des variations plus rapides que celui obtenu par de précédentes études. D'autre part, le champ magnétique obtenu pour les époques les plus récentes est raisonnablement similaire à celui construit à partir de mesures directes (satellites, observatoires, historiques) malgré des données beaucoup moins nombreuses et une répartition beaucoup moins homogène. Les modèles issus de cette étude offrent une alternative aux modèles existants régularisés, et pourront servir dans un objectif d'assimilation de données avec des modèles de la dynamique dans le noyau terrestre. / The aim of this thesis is to build stochastic models of the magnetic field for the last four millenia from archeomagnetic measurements. The sparse repartition of these data in space and time, and their associated large measurement and dating errors lead to an ill-posed problem. To determine the best solution, one needs to choose some prior information which consists usually on arbitrary regularizations in space and time. Instead, we use the temporal statistics of the geomagnetic field available from satellites, observatories and paleomagnetic measurements, and validated by numerical simulations, to define our prior information via auto-covariance functions. This bayesian method allows to get rid of arbitrary support functions, like splines, usually necessary to interpolate the model in time. The result consists in an ensemble of several possible realizations of the magnetic field. The ensemble dispersion represents the model uncertainties. We find that the methodology can be adapted to account for the age uncertainties and we use Markov Chain Monte Carlo to explore the possible dates of observations. This method improves the bootstrap method which gives the same weight to every draws of dates presenting very disparate probabilities. Each ensemble of realizations is then constructed from each selected model and the result is presented as a probability density function. The bayesian method together with the Markov Chain Monte Carlo provides regional time series with rapid variations compared to previous studies. We find that the possible values of geomagnetic field elements are not necessarily normally distributed. Another output of the model is better age estimates of archeological artefacts. The bayesian method has been used to build global models for which the axial dipole presents more rapid variations than for previous studies. Moreover, the obtained magnetic field displays reasonably similar behavior than models obtained from direct measurements (satellites, observatories, historical), despite very few data and sparser repartition. Models obtained from this study offer an alternative to published regularized models and can be used in a purpose of data assimilation together with dynamical models in the Earth's core.

L’électro-localisation est le nom donné aux capacités sensorielles de certains poissons électriques, vivant en eaux troubles, capables de détecter les perturbations électrostatiques dues à la présence d’objets dans leurs voisinages. Cette aptitude à interpréter un signal électrique pour se repérer dans l’espace ouvre l’importance perspectives, notamment dans le domaine de la robotique brio-inspiré. Mathématiquement, l’électrolocalisation est proche de la tomographie d’impédance électrique : il s’agit donc d’un problème inverse non linéaire, notoirement mal posé. Nous proposons dans cette thèse d’étudier des méthodes de reconstruction qui permettraient d’obtenir de manière robuste certaines caractéristiques de la forme des obstacles, plutôt que l’ensemble des détails de leurs géométries. Il s’agit donc d’étudier la stabilité de la partie observable des obstacles par rapport à des erreurs dans les mesures. / Electrolocation is the name given to the sensor ability for certain electric fish robots, which are able to detect electrostatic perturbations caused to the presence of some objects in their neighborhood. This ability to interpret an electrical signal to locate itself in space opens important perspectives, including in the field of biologically inspired robotics. Mathematically, electrolocation is linked to the electric impedance tomography: so it’s about a non-linear inverse problem, particularly ill-posed problem. We will, in this Phd, study some methods of reconstruction, which could be obtain robustly some characteristic of the obstacle’s shape, rather all of their geometry details. So, it’s about to study the stability between the observable part of the obstacles and the errors of measurements.

Title: Solution of inverse problem for a flow around an airfoil Author: Mgr. Jan Šimák Department: Department of Numerical Mathematics Supervisor: prof. RNDr. Miloslav Feistauer, DrSc., dr. h. c., Department of Numerical Mathematics Abstract: The method described in this thesis deals with a solution of an inverse problem for a flow around an airfoil. It can be used to design an airfoil shape according to a specified velocity or pressure distribution along the chord line. The method is based on searching for a fixed point of an operator, which combines an approximate inverse and direct operator. The approximate inverse operator, derived on the basis of the thin airfoil theory, assigns a corresponding shape to the specified distribution. The resulting shape is then constructed using the mean camber line and thickness function. The direct operator determines the pressure or velocity distribution on the airfoil surface. We can apply a fast, simplified model of potential flow solved using the Fredholm integral equation, or a slower but more accurate model of RANS equations with a k-omega turbulence model. The method is intended for a subsonic flow.

Ce travail de thèse apporte des éléments méthodologiques pour l'interprétation rapide de données électromagnétiques transitoires (TEM) aéroportées. Dans un premier temps, nous nous sommes intéressés au problème de l'inversion 1D qui est, à ce jour, un traitement standard. Nous nous sommes ensuite focalisés sur le problème de l'inversion 2D rapide. En nous basant sur les résultats de modélisations numériques, nous proposons un modèle empirique de sensibilité pour effectuer une inversion 2D rapide. Les tests effectués sur des données synthétiques et réelles s'avèrent très prometteurs. Ensuite, nous étudions analytiquement l'effet de la topographie. Pour un demi-espace homogène incliné, nous montrons que la direction des courants induits dans le sous-sol dépend de l'angle de l'interface air/sol. Nous proposons alors une méthode pour prendre en compte cet effet. Enfin, nous testons nos méthodes de traitement sur des données VTEM acquises dans le bassin de Franceville au Gabon. / This PhD work provides methodological tools for the fast interpretation of airborne transient electromagnetic data (ATEM). As our first step, we focused on the problem of 1D ground imaging, which is currently a standard treatment. Then, we focused on the fast 2D imaging problem. We proposed an empirical sensitivity function that allows fast 2D inversion. The results of the 2D empirical inversion when applied on both synthetic and real data sets are very promising. In the third part, we analyzed the effect of topography on the airborne EM data. For an inclined homogeneous half-space, we show that the direction of the ground induced current is parallel to the surface air/ground. Then, we proposed a method to take into account this effect. Finally, the tools developed during this PhD have been applied to a real VTEM data set acquired over the basin of Franceville, in Gabon.

A distribuição de massa molar (DMM) e seus parâmetros são de fundamental importância na caracterização dos polímeros. Por este motivo, o desenvolvimento de técnicas que permitam a determinação da DMM de forma mais rápida e a menor custo é de grande importância prática. Os principais objetivos deste trabalho foram a implementação de alguns dos modelos baseados da teoria da reptação dupla propostos na literatura para descrever o mecanismo de relaxação das cadeias poliméricas, a avaliação dessas implementações e a análise de dois passos fundamentais na obtenção da DMM a partir de dados reológicos que são a metodologia de cálculo do espectro de relaxação baseado no modelo de Maxwell e a estratégia para a avaliação numérica das integrais que aparecem nos modelos de relaxação. Foi resolvido o problema denominado problema inverso, ou seja, a determinação da DMM a partir de dados reológicos usando um modelo de relaxação especificado e uma função de distribuição imposta. Foi usada a função Exponencial Generalizada (GEX) para representar a probabilidade de distribuição, sendo consideradas duas abordagens: i) cálculo explícito do espectro de relaxação e ii) aproximações paramétricas de Schwarzl, que evitam a necessidade do cálculo explícito do espectro de relaxação. A metodologia de determinação da DMM foi aplicada para amostras de polietileno e foram estimadas distribuições com boa representação dos dados experimentais do GPC, ao considerarem-se amostras com polidispersões inferiores a 10. Com relação a metodologia de cálculo do espectro de relaxação, foi realizado um estudo comparativo da aplicação de espectros de relaxação discreto e contínuo, com o objetivo de estabelecer critérios para especificação do número ótimo de modos de Maxwell a serem considerados. Ao efetuar-se a comparação entre as técnicas, verificou-se o espectro discreto apresenta como um sistema melhor condicionado, permitindo assim obter maior confiabilidade dos parâmetros estimados. Também é proposta uma modificação da metodologia de determinação da DMM, em que é aplicada a quadratura de Gauss-Hermite para a resolução numérica da integral dos modelos de relaxação. / The molecular weight distribution (MWD) and its parameters are of the fundamental importance in the characterization of polymers. Therefore, the development of techniques for faster and less time consuming determination of the MWD is of great practical relevance. The goals of this work were the implementation of some of the relaxation models from double reptation theory proposed in the literature, the evaluation of these implementations and the analysis of two key points in the recovery of the MWD from rheological data which are the methodology for calculation of the relaxation spectrum based on the Maxwell model and the numeric strategy for the evaluation of the integrals appearing in the relaxation models. The inverse problem, i.e., the determination of the MWD from rheological data using a specified relaxation model and an imposed distribution function, was solved. In the analysis of the inverse problem, the Generalized Exponential (GEX) was used as distribution function and two approaches were considered: i) explicit calculation of the relaxation spectrum and ii) use of the parametric method proposed by Schwarzl to avoid the explicit calculation of the relaxation spectrum. In the test of commercial samples of polyethylene with polidispersity less than 10, the application of this methodology led to MWD curves which provided good fit of the experimental SEC data. Regarding the methodology for calculation of the relaxation spectrum, a comparison between the performance of discrete and continuous relaxation spectrum was performed and some possible a criteria to determine the appropriate number of relaxation modes of Maxwell to be used were evaluated. It was found that the technique of discrete spectrum leads to better conditioned systems and, consequently, greater reliability of the estimated parameters. With relation to the numeric strategy for the evaluation of the integrals appearing in the relaxation models, the use of Gauss-Hermite quadrature using a new change of variables was proposed.

Abstract This dissertation is a combination of four articles on the topic of scattering problems for a biharmonic operator. The operator of interest has two coefficients which may be complex-valued and singular. Each of the articles concerns a different aspect of the problem. Namely, the first article discusses the direct scattering problem in higher dimensions and culminates in a proof of Saito's formula, which yields a uniqueness result for the inverse scattering problem. The second paper is about a backscattering problem in two and three dimensions. We prove that the inverse Born approximation can be used to recover the singularities in the coefficients of the operator. The third article fills in an answer to the question about recovering the complex-valued coefficients in three dimensions that was left open in the second article. The final article studies the inverse scattering problem on the line for a quasi-linear operator. / Tiivistelmä Väitöskirjatyö koostuu neljästä artikkelista, jotka käsittelevät sirontaongelmia biharmoniselle operaattorille. Työn kohteena olevalla operaattorilla on kaksi kerrointa, jotka voivat olla kompleksiarvoisia ja singulaarisia. Kukin artikkeli käsittelee sirontaongelmaa eri näkökulmasta. Ensimmäinen artikkeli koostuu pääasiassa suorasta sirontateoriasta korkeammissa ulottuvuuksissa huipentuen lopulta Saiton kaavan todistukseen, jonka seurauksena saadaan yksikäsitteisyystulos käänteiselle sirontaongelmalle. Toisen artikkelin aiheena on takaisinsirontaongelma kahdessa ja kolmessa ulottuvuudessa. Todistamme, että käänteistä Bornin approksimaatiota voidaan käyttää paikantamaan kertoimien mahdolliset singulariteetit. Kolmas artikkeli vastaa toisessa artikkelissa avoimeksi jääneeseen kysymykseen kompleksiarvoisien kertoimien rekonstruoimisesta kolmessa ulottuvuudessa. Viimeisessä artikkelissa tutkitaan käänteistä sirontaongelmaa kvasilineaariselle operaattorille yhdessä ulottuvuudessa.