We consider the problem of structural optimization which has many important applications in the engineering sciences. The goal is to find an optimal distribution of the material within a certain volume that will minimize the mechanical and/or thermal compliance of the structure. The physical system is governed by the standard models of elasticity and heat transfer expressed in terms of boundary-value problems for elliptic systems of partial differential equations (PDEs). The structural optimization problem is then posed as a suitably constrained PDE optimization problem, which can be solved numerically using a gradient approach. As a main contribution to the thesis, we derive expressions for gradients (sensitivities) of different objective functionals. This is done in both the continuous and discrete setting using the Riesz representation theorem and adjoint analysis. The sensitivities derived in this way are then tested computationally using simple minimization algorithms and some standard two-dimensional test problems. / Thesis / Master of Science (MSc)

In 1909, F. Riesz succeeded in giving an integral represntation for continuous linear functionals on C[0,1]. Although other authors, notably Hadamard and Frechet, had given representations for continuous linear functionals on C[0,1], their results lacked the clarity, elegance, and some of the substance (uniqueness) of Riesz's theorem. Subsequently, the integral representation of continuous linear functionals has been known as the Riesz Representation Theorem. In this paper, three different proofs of the Riesz Representation Theorem are presented. The first approach uses the denseness of the Bernstein polynomials in C[0,1] along with results of Helly to write the continuous linear functionals as Stieltjes integrals. The second approach makes use of the Hahn-Banach Theorem in order to write the functional as an integral. The paper concludes with a detailed presentation of a Daniell integral development of the Riesz Representation Theorem.

A well known result in probability is that convergence almost surely (a.s.) of a sequence of random elements implies weak convergence of their laws. The Ukrainian mathematician Anatoliy Volodymyrovych Skorohod proved the lemma known as Skorohods a.s. representation Theorem, a partial converse of this result. In this work we discuss the notion of continuous representations, which allows us to provide generalizations of Skorohods Theorem. In Chapter 2, we explore Blackwell and Dubinss extension [3] and Ferniques extension [10]. In Chapter 3 we present Cortissozs result [5], a variant of Skorokhods Theorem. It is shown that given a continuous path inM(S) it can be associated a continuous path with fixed endpoints in the space of S-valued random elements on a nonatomic probability space, endowed with the topology of convergence in probability. In Chapter 4 we modify Blackwell and Dubins representation for particular cases of S, such as certain subsets of R or R^n. / Mathematics

No description available.

We look from a new point of view at the definition and basic properties of the Lebesgue measure and integral on Euclidean spaces, on abstract spaces, and on locally compact Hausdorff spaces. We use mini sums to give all of them a unified treatment that is more efficient than the standard ones. We also give Fubini's theorem a proof that is nicer and uses much lighter technical baggage than the usual treatments.

If T:C(H,X)-->Y is a bounded linear operator then there exists a unique weakly regular finitely additive set function m:-->L(X,Y**) so that T(f) = ∫Hfdm. In this paper, bounded linear operators on C(H,X) are studied in terms the measure given by this representation theorem. The first chapter provides a brief history of representation theorems of these classes of operators. In the second chapter the represenation theorem used in the remainder of the paper is presented. If T is a weakly compact operator on C(H,X) with representing measure m, then m(A) is a weakly compact operator for every Borel set A. Furthermore, m is strongly bounded. Analogous statements may be made for many interesting classes of operators. In chapter III, two classes of operators, weakly precompact and QSP, are studied. Examples are provided to show that if T is weakly precompact (QSP) then m(A) need not be weakly precompact (QSP), for every Borel set A. In addition, it will be shown that weakly precompact and GSP operators need not have strongly bounded representing measures. Sufficient conditions are provided which guarantee that a weakly precompact (QSP) operator has weakly precompact (QSP) values. A sufficient condition for a weakly precomact operator to be strongly bounded is given. In chapter IV, weakly precompact subsets of L1(μ,X) are examined. For a Banach space X whose dual has the Radon-Nikodym property, it is shown that the weakly precompact subsets of L1(μ,X) are exactly the uniformly integrable subsets of L1(μ,X). Furthermore, it is shown that this characterization does not hold in Banach spaces X for which X* does not have the weak Radon-Nikodym property.

This work presents a systematic investigation of two topics. One is in stochastic dynamic general equilibrium. It incorporates private information into dynamic general equilibrium framework. An existence of competitive equilibrium is established. Quantitative analysis is provided for health insurance problem. The other topic is in decision problems under ambiguity. Lack of precise information regarding a decision problem is represented by a set of probabilities. Descriptive richness of the set of probabilities is defi ned. It is used to generalize Skorohod's theorem to sets of probabilities. The latter is used to show the constancy of the coefficient in alpha-maximin multiple prior preferences. Examples illustrate: the implications of this representation; and the restrictions arising from the failure of descriptive richness. / text

We look at three distinct topics in analysis. In the first we give a direct and easy proof that the usual Newton-Leibniz rule implies the fundamental theorem of algebra that any nonconstant complex polynomial of one complex variable has a complex root. Next, we look at the Riesz representation theorem and show that the Riesz representing measure often can be given in the form of mini sums just like in the case of the usual Lebesgue measure on a cube. Lastly, we look at the idea of holomorphic domination and use it to define a class of complex Banach manifolds that is similar in nature and definition to the class of Stein manifolds.

Dados um espaço de medida (X;A;m) e números reais p,q>1 com 1/p+1/q=1, o Teorema de Representação de Riesz afirma que Lq(X;A;m) é o dual topológico de Lp(X;A;m) e que Loo(X;A; m) é o dual topológico de L1(X;A;m) se o espaço (X;A;m) for sigma-finito. Observamos que a sigma-finitude de (X;A;m) é condição suficiente mas não necessária para que Loo(X;A;m) seja o dual de L1(X;A;m). Os contra-exemplos tipicamente apresentados para essa última identificação são \"triviais\", no sentido de que desaparecem se \"consertarmos\" a medida , transformando-a numa medida perfeita. Neste trabalho apresentamos condições sufcientes mais fracas que sigma-finitude a fim de que Loo(X;A;m) e o dual de L1(X;A;m) possam ser isometricamente identificados. Além disso, introduzimos um invariante cardinal para espaços de medida que chamaremos a dimensão do espaço e mostramos que se o espaço (X;A;m) for de medida perfeita e tiver dimensão menor ou igual à cardinalidade do continuum então uma condição necessária e suficiente para Loo(X;A;m) seja o dual de L1(X;A;m) é que X admita uma decomposição. / Given a measure space (X;A;m) and real numbers p,q>1 with 1/p+1/q=1, the Riesz Representation Theorem states that Lq(X;A;m) is the topological dual space of Lp(X;A;m) and that Loo(X;A; m) is the topological dual space of L1(X;A;m) if (X;A; m) is sigma-finite. We observe that the sigma-finiteness of (X;A;m) is a suficient but not necessary condition for Loo(X;A;m) to be the dual of L1(X;A;m). The counter-examples that are typically presented for Loo(X;A;m) = L1(X;A;m)* are \"trivial\", in the sense that they vanish if we fix the measure , making it into a perfect measure. In this work we present suficient conditions weaker than sigma-finiteness in order that Loo(X;A; m) and L1(X;A;m)* can be isometrically identified. Moreover, we introduce a cardinal invariant for measure spaces which we call the dimension of the space and we show that if the space (X;A;m) has perfect measure and dimension less than or equal to the cardinal of the continuum then a necessary and suficient condition for Loo(X;A;m) = L1(X;A;m)* is that X admits a decomposition.

Inverse problems occur in a wide variety of applications and are an active area of research in many disciplines. We consider inverse problems for a broad class of nonlinear systems of partial differential equations (PDEs). We develop collage-based approaches for solving inverse problems for nonlinear PDEs of elliptic, parabolic and hyperbolic type. The original collage method for solving inverse problems was developed in [29] with broad application, in particular to ordinary differential equations (ODEs). Using a consequence of Banach’s fixed point theorem, the collage theorem, one can bound the approximation error above by the so-called collage distance, which is more readily minimizable. By minimizing the collage distance the approximation error can be controlled. In the case of nonlinear PDEs we consider the weak formulation of the PDE and make use of the nonlinear Lax-Milgram representation theorem and Galerkin approximation theory in order to develop a similar upper-bound on the approximation error. Supporting background theory, including weak solution theory,is presented and example problems are solved for each type of PDE to showcase the methods in practice. Numerical techniques and considerations are discussed and results are presented. To demonstrate the practical applicability of this work, we study two real-world applications. First, we investigate a model for the migration of three fish species through floodplain waters. A development of the mathematical model is presented and a collage-based method is applied to this model to recover the diffusion parameters. Theoretical and numerical particulars are discussed and results are presented. Finally, we investigate a model for the “Gao beam”, a nonlinear beam model that incorporates the possibility of buckling. The mathematical model is developed and the weak formulation is discussed. An inverse problem that seeks the flexural rigidity of the beam is solved and results are presented. Finally, we discuss avenues of future research arising from this work. / Natural Sciences and Engineering Research Council of Canada, Department of Mathematics & Statistics