We develop a theory concerning the reconstruction of multivariate exponential sums first over arbitrary fields and then consider the special cases of multivariate exponential sums over the fields of real and complex numbers.

This thesis describes possible applications of Prony's method in timing analysis of digital circuits. Such applications include predicting the future shape of the waveform in DTA(Dynamic Timing Analysis) and delay look-up table in STA(Static Timing Analysis). Given some equally spaced output values, the traditional Prony's method can be used to extract poles and residues of a linear system, i.e. to characterize a waveform using an exponential function. In this thesis, not only values but also equally spaced derivatives are tested. Still using same idea of the traditional Prony's method, poles and residues can also be extracted with those values and derivatives. The resultant poles and residues will be used to predict the output waveform in DTA analysis. The benefits brought by the using of derivatives include less simulation steps and less CPU time consuming than the regular constant step simulation. As a matter of fact, the Prony's method can precisely approximate a complicated waveform. Such property can be applied for STA analysis. The Prony's approximation can be used to precisely record an output waveform, which is used as an entry of the look-up table of STA. Since the accuracy of STA analysis relies on the accuracy of the input and output waveform in the look-up table, the accuracy of the Prony's approach is promising.

This thesis describes possible applications of Prony's method in timing analysis of digital circuits. Such applications include predicting the future shape of the waveform in DTA(Dynamic Timing Analysis) and delay look-up table in STA(Static Timing Analysis). Given some equally spaced output values, the traditional Prony's method can be used to extract poles and residues of a linear system, i.e. to characterize a waveform using an exponential function. In this thesis, not only values but also equally spaced derivatives are tested. Still using same idea of the traditional Prony's method, poles and residues can also be extracted with those values and derivatives. The resultant poles and residues will be used to predict the output waveform in DTA analysis. The benefits brought by the using of derivatives include less simulation steps and less CPU time consuming than the regular constant step simulation. As a matter of fact, the Prony's method can precisely approximate a complicated waveform. Such property can be applied for STA analysis. The Prony's approximation can be used to precisely record an output waveform, which is used as an entry of the look-up table of STA. Since the accuracy of STA analysis relies on the accuracy of the input and output waveform in the look-up table, the accuracy of the Prony's approach is promising.

Linearized stability analysis methodologies, system identification algorithms and optimal control approaches that are applicable to large scale, flexible multibody dynamic systems are presented in this thesis. For stability analysis, two classes of closely related algorithms based on a partial Floquet approach and on an autoregressive approach, respectively, are presented in a common framework that underlines their similarity and their relationship to other methods. The robustness of the proposed approach is improved by using optimized signals that are derived from the proper orthogonal modes of the system. Finally, a signal synthesis procedure based on the identified frequencies and damping rates is shown to be an important tool for assessing the accuracy of the identified parameters; furthermore, it provides a means of resolving the frequency indeterminacy associated with the eigenvalues of the transition matrix for periodic systems. For system identification, a robust algorithm is developed to construct subspace plant models. This algorithm uniquely combines the methods of minimum realization and subspace identification. It bypasses the computation of Markov parameters because the free impulse response of the system can be directly computed in the present computational environment. Minimum realization concepts were applied to identify the stability and output matrices. On the other hand, subspace identification algorithms construct a state space plant model of linear system by using computationally expensive oblique matrix projection operations. The proposed algorithm avoids this burden by computing the Kalman filter gain matrix and model dependency on external inputs in a small sized subspace. Balanced model truncation and similarity transformation form the theoretical foundation of proposed algorithm. Finally, a forward innovation model is constructed and estimates the input-output behavior of the system within a specified level of accuracy. The proposed system identification algorithms are computationally inexpensive and consist of purely post processing steps that can be used with any multi-physics computational tool or with experimental data. Optimal control methodologies that are applicable to comprehensive large-scale flexible multibody systems are presented. A classical linear quadratic Gaussian controller is designed, including subspace plant identification, the evaluation of linear quadratic regulator feedback gain and Kalman filter gain matrices and online control implementation.

This thesis describes possible applications of Prony's method in timing analysis of digital circuits. Such applications include predicting the future shape of the waveform in DTA(Dynamic Timing Analysis) and delay look-up table in STA(Static Timing Analysis). Given some equally spaced output values, the traditional Prony's method can be used to extract poles and residues of a linear system, i.e. to characterize a waveform using an exponential function. In this thesis, not only values but also equally spaced derivatives are tested. Still using same idea of the traditional Prony's method, poles and residues can also be extracted with those values and derivatives. The resultant poles and residues will be used to predict the output waveform in DTA analysis. The benefits brought by the using of derivatives include less simulation steps and less CPU time consuming than the regular constant step simulation. As a matter of fact, the Prony's method can precisely approximate a complicated waveform. Such property can be applied for STA analysis. The Prony's approximation can be used to precisely record an output waveform, which is used as an entry of the look-up table of STA. Since the accuracy of STA analysis relies on the accuracy of the input and output waveform in the look-up table, the accuracy of the Prony's approach is promising.

This thesis describes possible applications of Prony's method in timing analysis of digital circuits. Such applications include predicting the future shape of the waveform in DTA(Dynamic Timing Analysis) and delay look-up table in STA(Static Timing Analysis). Given some equally spaced output values, the traditional Prony's method can be used to extract poles and residues of a linear system, i.e. to characterize a waveform using an exponential function. In this thesis, not only values but also equally spaced derivatives are tested. Still using same idea of the traditional Prony's method, poles and residues can also be extracted with those values and derivatives. The resultant poles and residues will be used to predict the output waveform in DTA analysis. The benefits brought by the using of derivatives include less simulation steps and less CPU time consuming than the regular constant step simulation. As a matter of fact, the Prony's method can precisely approximate a complicated waveform. Such property can be applied for STA analysis. The Prony's approximation can be used to precisely record an output waveform, which is used as an entry of the look-up table of STA. Since the accuracy of STA analysis relies on the accuracy of the input and output waveform in the look-up table, the accuracy of the Prony's approach is promising.

The thesis studies truncated moment problems and related reconstruction techniques. It transfers the main aspects of Prony's method from finitely-supported measures to the classes of signed or non-negative measures supported on algebraic varieties of any dimension. The Zariski closure of the support of these measures is shown to be determined by finitely many moments and can be computed from the kernel of moment matrices. Moreover, several reconstruction algorithms are developed which are based on the computation of generalized eigenvalues and allow to recover the components of mixtures of such measures.

Cette thèse porte sur la reconstruction de formes linéaires sur l'anneau des polynômes dans le cas multivarié et ses applications. Nous proposons des outils théoriques et algorithmiques permettant de résoudre des problèmes liés aux moments : la reconstruction de polytopes convexes à partir de leurs moments et la recherche de cubatures. L'algorithme numérique proposé pour reconstruire des polytopes utilise des méthodes numériques utilisées précédemment pour le cas des polygones, ainsi que les identités de Brion reliant moments directionnels et sommets projetés. Un polyèdre à 57 sommets - la coupe d'un diamant - est ainsi reconstruit. Pour la recherche de cubatures, nous adaptons la méthode de Prony univariée en une méthode multivariée à l'aide des opérateurs de Hankel. Un problème de complétion de matrices est aussi résolu grâce au théorème d'extension plate de Curto-Fialkow. Nous expliquons ainsi la recherche de cubatures à l'aide des matrices de moments, connue dans la littérature. La symétrie, qui est ici un élément naturel, réduit la complexité algorithmique. Nous prouvons qu'une diagonalisation par blocs des matrices concernées est alors possible. De ces blocs et à l'aide de la matrice de multiplicités d'un groupe fini, des conditions nécessaires à l'existence de cubatures sont obtenues. Pour une mesure, un degré et un nombre de nœuds donnés, notre algorithme certifie tout d'abord l'existence de cubatures et ensuite calcule ses poids et nœuds. De nouvelles cubatures ont ainsi été trouvées : soit en complétant celles connues pour une mesure et un degré donnés, soit en ajoutant des cubatures de degrés supérieurs pour une mesure donnée. / This thesis deals with the reconstruction of linear forms on the polynomial ring and its applications. We propose theoretical and algorithmic tools to solve multivariate moment problems: the reconstruction of convex polytopes from their moments (shape-from-moments) and the search for cubatures. The numerical algorithm we propose to reconstruct polytopes uses numerical methods previously known in the case of polygons, and also Brion's identities that relate directional moments and projected vertices. A polyhedron with 57 vertices – a diamond cut – is thus reconstructed. Concerning the search for cubatures, we adapt the univariate Prony's method into a multivariate method thanks to Hankel operators. A matrix completion problem is then solved with a basis-free version of Curto-Fialkow's flat extension theorem. We explain thus the moment matrix approach to cubatures, known in the litterature. Symmetry is here a natural ingredient and reduces the algorithmic complexity. We show that a block diagonalisation of the involved matrices is possible. Those blocs and the matrix of multiplicities of a finite group provide necessary conditions on the existence of cubatures. Given a measure, a degree and a number of nodes, our algorithm first certify the existence of cubatures and then compute the weights and nodes. New cubatures have been found: either by completing the ones known for a given measure and degree, or by adding cubatures with a higher degree for a given measure.

Radars, as the name suggests, were traditionally used for Radio Detection and Ranging. Nevertheless, advances in high resolution electromagnetic simulations, Ultra Wide-Band sources, signal processing and computer technologies have resulted in a possible perception of radars as sensors for target discrimination. In this thesis, the feasibility of discrimination between targets even with minor variations in structure and material composition on the basis of radar echoes is effectively demonstrated. It is well-known that the echoes from any target are affected by its natural frequencies which are dependent only on the shape and material composition of the target, and independent of the aspect angle or the incident waveform. The E-pulse technique is based on the fact that incident waveforms can be designed that uniquely annihilate the echoes from chosen regions of a target, and forms the basis of the method of discrimination proposed in this thesis. Earlier methods reported in the literature, effectively discriminated only between different classes of targets with substantial variations in the overall dimensions of the body. Discrimination of targets of the same class with a minor structural modification or with a material coating on specific areas was rather difficult. This thesis attempts and successfully validates a method which comprehensively addresses this problem. The key idea of this method is to use the higher frequency resonances (which characterise the finer details of a target) in the E-pulse technique. An obviously important aspect of target discrimination is therefore that of precisely estimating the natural frequencies for each target and understanding the changes in these frequencies, and their associations with the changes in structure and material composition. Current approaches to determine these frequencies are either based In the time or frequency domains. While the latter approach comprises the computation of the roots of a related determinantal equation, in the time domain, the natural frequencies are extracted from the response of a target to an impulse. Such a response can either be generated from actual experiments or by simulating the scattering response using Computational Electromagnetic (CEM) techniques. In this work, the impulse response is obtained from the frequency response of the scatterers in the frequency range of interest. Since no single CEM technique can effectively cover the entire range of frequencies needed for the E-Pulse synthesis. The Method of Moments and Physical Optics have been used for low and high frequency scattering respectively. The results obtained using the latter technique are validated by comparing with those obtained using Method of Moments at the transition frequencies and Geometrical Theory of Diffraction (GTD). The natural frequencies (i.e., poles of a corresponding transfer function) are extracted from the impulse response using Prony's algorithm. One of the parameters in this method is the number of such poles (i.e.. the order of the transfer function) present in the response, and the accuracy of the computed pole values depends on this assumed order. Here, the Hankel singular values of a transfer function is used to estimate the number of poles. This in turn implies that a specific norm of the error between a transfer function corresponding to the frequency response generated earlier, and a transfer function with an assumed order obtained using Prony's method is minimised. In the thesis, a wide range of target shapes are considered for purposes of illustration: wires, cylinders, spheres, plates and complex bodies such as aircraft, and the discrimination capability is demonstrated by introducing minor perturbations in their shape and/or material composition. .The following cases are considered here: (a) Wires: Conducting wires with a protrusion in one segment; conducting wire from another coated with a dielectric in a segment, (b) Cylinders: Conducting cylinders with one perturbed; conducting cylinders with a portion scrapped off in the middle, (c) Plates: Conducting plates with a elongation on one comer; conducting plate with another one with a hole in the centre, (d) Spheres: Conducting spheres with different radii; conducting spheres with Radar Absorbing Material coated spheres with different coating thickness; conducting spheres with chiral coated spheres with varying coating thickness, (e) Aircraft: Canonical model of MiG-29 aircraft from a similar one with stores placed under the wing.