In this thesis, we consider the question of describing the eigenvalues of a distinguished family of invariant differential operators associated to a Lie superalgebra g and a g-module W, called the "Capelli basis", via evaluation of certain classes of supersymmetric functions, called the interpolation super Jack polynomials. Finding the eigenvalues of the Capelli basis is referred to the Capelli Eigenvalue Problem. The eigenvalue formula depends on the chosen parametrization of the highest weight vectors in the decomposition of the superpolynomial algebra P(W), and consequently on the choice of a Borel subalgebra. In this thesis, we give a solution for each conjugacy class of Borel subalgebras, which we call a refined solution to the Capelli Eigenvalue Problem. Given the pair (g, W), we investigate the formulae for the eigenvalues of the Capelli operators associated to the completely reducible and multiplicity-free modules for two cases: diagonal and symmetric cases. In the former case, we show that we can express the eigenvalue of the Capelli operator on the irreducible component of the multiplicity-free decomposition of P(W) as a polynomial function of the b-highest weight of the irreducible component for any Borel subalgebra b. In the latter case, we show with a concrete counterexample that we cannot expect the results to be as strong as in the first case for all Borel subalgebras. We then express the eigenvalue of the Capelli operator on the irreducible component of the multiplicity-free decomposition of P(W) as a polynomial function of a piecewise affine map on the span of b-highest weights of the irreducible submodules of P(W), with respect to different decreasing Borel subalgebras b.

This thesis demonstrates an efficient parallel method of solving the generalized eigenvalue problem, KΦ = M ΦΛ, where K is symmetric and M is symmetric positive-definite, by first converting it to a standard eigenvalue problem, solving the standard eigenvalue problem, and back-transforming the results. An abstraction for parallel dense linear algebra is introduced along with a new algorithm for forming A := U⁻ᵀ K U⁻¹ , where U is the Cholesky factor of M , that is up to twice as fast as the ScaLAPACK implementation. Additionally, large improvements over the PBLAS implementations of general matrix-matrix multiplication and triangular solves with many right-hand sides are shown. Significant performance gains are also demonstrated for Cholesky factorizations, and a case is made for using 2D-cyclic distributions with a distribution blocksize of one. / text

No description available.

This Thesis contains contributions in two different but closely related subfields of Scientific and Parallel Computing which arise in the context of various eigenvalue problems: periodic and parallel eigenvalue reordering and parallel algorithms for Sylvestertype matrix equations with applications in condition estimation. Many real world phenomena behave periodically, e.g., helicopter rotors, revolving satellites and dynamic systems corresponding to natural processes, like the water flow in a system of connected lakes, and can be described in terms of periodic eigenvalue problems. Typically, eigenvalues and invariant subspaces (or, specifically, eigenvectors) to certain periodic matrix products are of interest and have direct physical interpretations. The eigenvalues of a matrix product can be computed without forming the product explicitly via variants of the periodic Schur decomposition. In the first part of the Thesis, we propose direct methods for eigenvalue reordering in the periodic standard and generalized real Schur forms which extend earlier work on the standard and generalized eigenvalue problems. The core step of the methods consists of solving periodic Sylvester-type equations to high accuracy. Periodic eigenvalue reordering is vital in the computation of periodic eigenspaces corresponding to specified spectra. The proposed direct reordering methods rely on orthogonal transformations and can be generalized to more general periodic matrix products where the factors have varying dimensions and ±1 exponents of arbitrary order. In the second part, we consider Sylvester-type matrix equations, like the continuoustime Sylvester equation AX −XB =C, where A of size m×m, B of size n×n, and C of size m×n are general matrices with real entries, which have applications in many areas. Examples include eigenvalue problems and condition estimation, and several problems in control system design and analysis. The parallel algorithms presented are based on the well-known Bartels–Stewart’s method and extend earlier work on triangular Sylvester-type matrix equations resulting in a novel software library SCASY. The parallel library provides robust and scalable software for solving 44 sign and transpose variants of eight common Sylvester-type matrix equations. SCASY also includes a parallel condition estimator associated with each matrix equation. In the last part of the Thesis, we propose parallel variants of the direct eigenvalue reordering method for the standard and generalized real Schur forms. Together with the existing and future parallel implementations of the non-symmetric QR/QZ algorithms and the parallel Sylvester solvers presented in the Thesis, the developed software can be used for parallel computation of invariant and deflating subspaces corresponding to specified spectra and associated reciprocal condition number estimates.

There are numerous applications of mathematics, specifically spectral graph theory, within the sciences and many other fields. This paper is an exploration of recent applications of spectral graph theory, including the fields of chemistry, biology, and graph coloring. Topics such as the isomers of alkanes, the importance of eigenvalues in protein structures, and the aid that the spectra of a graph provides when coloring a graph are covered, as well as others.The key definitions and properties of graph theory are introduced. Important aspects of graphs, such as the walks and the adjacency matrix are explored. In addition, bipartite graphs are discussed along with properties that apply strictly to bipartite graphs. The main focus is on the characteristic polynomial and the eigenvalues that it produces, because most of the applications involve specific eigenvalues. For example, if isomers are organized according to their eigenvalues, a pattern comes to light. There is a parallel between the size of the eigenvalue (in comparison to the other eigenvalues) and the maximum degree of the graph. The maximum degree of the graph tells us the most carbon atoms attached to any given carbon atom within the structure. The Laplacian matrix and many of its properties are discussed at length, including the classical Matrix Tree Theorem and Cayley's Tree Theorem. Also, an alternative approach to defining the Laplacian is explored and compared to the traditional Laplacian.

Nous décrivons un nouvelle famille d'exemples d'hypersurfaces de la sphère satisfaisant le cas d'égalité de la majoration extrinsèque de C. Bär de la plus petite valeur propre de l'opérateur de Dirac. / We describe a new family of examples of hypersurfaces in the sphere satisfying the limitingcase in C. Bär's extrinsic upper bound for the smallest eigenvalue of the Dirac operator.

With the development of consumer electronics, the circuitry structure become more complex, For this reason, it might cause numerical errors to be cumulated in the simulation using the numerical electromagnetic algorithm, and result in simulated divergence or error. The two reasons of numerical error are passivity and causality, which priginate from the defect in the numerical calculation. In this thesis, for this problem, investigate the numerical compensation method for passivity, The occurrence of passive will make the frequency point of power is negative, this will makes the system divergence, Improve this problem, passivity verification and enforcement by eigenvalue in the Y-parameter, in the S-parameter by the singular value, causality conditions must be match with the imaginary part and the real part relationship, such as the Hilbert transform or the Kramer-Kronig relation, can be used to make causal verification and enforcement. Through some numerical methods, used simulation software such as: HFSS, ADS simulation of the microwave circuit model extraction, modified singular value, eigenvalue, and reached to reduce the numerical error, let it satisfy the convergence and avoid incorrect results, and minimize the impact of the initial data, does not change the characteristics of the original module, but also to solve the passive and the issue of causality.

The inverse spectral problem is the problem of understanding the potential function of the Sturm-Liouville operator from the set of eigenvalues plus some additional spectral data. The theory of transformation operators, first introduced by Marchenko, and then reinforced by Gelfand and Levitan, is a powerful method to deal with the different stages of the inverse spectral problem: uniqueness, reconstruction, stability and existence. In this thesis, we shall give a survey on the theory of transformation operators. In essence, the theory says that the transformation operator $X$ mapping the solution of a Sturm-Liouville operator $varphi$ to the solution of a Sturm-Liouville operator, can be written as $$Xvarphi=varphi(x)+int_{0}^{x}K(x,t)varphi(t)dt,$$ where the kernel $K$ satisfies the Goursat problem $$K_{xx}-K_{tt}-(q(x)-q_{0}(t))K=0$$ plus some initial boundary conditions. Furthermore, $K$ is related by a function $F$ defined by the spectral data ${(lambda_{n},alpha_{n})}$ where $alpha_{n}=(int_{0}^{pi}|varphi_{n}(t)|^{2})^{frac{1}{2}}$ through the famous Gelfand-Levitan equation $$K(x,y)+F(x,y)+int_{o}^{x}K(x,t)F(t,y)dt=0.$$ Furthermore, all the above relations are bilateral, that is $$qLeftrightarrow KLeftrightarrow FLeftarrow {(lambda_{n},alpha_{n})}.$$ hspace*{0.25in}We shall give a concise account of the above theory, which involves Riesz basis and order of entire functions. Then, we also report on some recent applications on the uniqueness result of the inverse spectral problem.

In this thesis we explore the relationship between aeroelastic flutter and multiparameter spectral theory. We first introduce the basic concept of the relationship between these two fields in abstract terms. Then we expand on this initial concept, using it to devise visualisation methods and a wide variety of solvers for flutter problems. We assess these solvers, applying them to real-life aeroelastic systems and measuring their performance. We then discuss and devise methods for improving these solvers. All our conclusions are supported by a variety of evidence from numerical experiments. Finally, we assess all of our methods, providing recommendations as to their use and future development. We do achieve several things in this thesis which have not been achieved before. Firstly, we solved a non-trivial flutter problem with a direct solver. The only direct solvers that have previously been presented are those that arise from classical flutter analysis, which applies only to very simple systems. Secondly, and as an extension of this first point, we solved a system with Theodorsen aerodynamics (approximated by a highly accurately) with a direct solver. This was achieved in an industrially competitive time (0.2s). This has never before been achieved. Thirdly, we solved an unstructured multiparameter eigenvalue problem. Unstructured problems have not been considered before, even in theoretical literature. This result is thus of significance both for multiparameter spectral theory and aeroelasticity. However, the single most important contribution of this thesis is the opening of a whole new field of study which stretches beyond aeroelasticity and into other industries: the treatment of instability problems using multiparameter methods. This field of research is wide and untrodden, and has the potential to change the way we analyse instability across many industries.

Using numerical simulations and combinatorics, this dissertation focuses on connections between random matrix theory and graph theory.We examine the adjacency matrices of three-regular graphs representing one-face maps. Numerical studies have revealed that the limiting eigenvalue statistics of these matrices are the same as those of much larger, and more widely studied classes of random matrices. In particular, the eigenvalue density is described by the McKay density formula, and the distribution of scaled eigenvalue spacings appears to be that of the Gaussian Orthogonal Ensemble (GOE).A natural question is whether the eigenvalue statistics depend on the genus of the underlying map. We present an algorithm for generating random three-regular graphs representing genus zero one-face maps. Our numerical studies of these three-regular graphs have revealed that their eigenvalue statistics are strikingly different from those of three-regular graphs representing maps of higher genus. While our results indicate that there is a limiting eigenvalue density formula in the genus zero case, it is not described by any established density function. Furthermore, the scaled eigenvalue spacings appear to be described by the exponential distribution function, not the GOE spacing distribution.The embedded graph of a genus zero one-face map is a planar tree, and there is a correlation between its vertices and the primitive cycles of the associated three-regular graph. The second half of this dissertation examines the structure of these embedded planar trees. In particular, we show how the Dyck path representation can be used to recast questions about the probabilistic structure of random planar trees into straightforward counting problems. Using this Dyck path approach, we find:1. the expected number of degree k vertices adjacent to j degree d vertices in a random planar tree, 2. the structure of the planar tree's adjacency matrix under a natural labeling of the vertices, and 3. an explanation for the existence of eigenvalues with multiplicity greater than one in the tree's spectrum.