Recursion is a fundamental tool of mathematics used to define, construct, and analyze mathematical objects. This work employs induction, sieving, inversion, and other recursive methods to solve a variety of problems in the areas of algebraic number theory, topological and combinatorial graph theory, and analytic probability and statistics. A common theme of recursively defined functions, weighted sums, and cross-referencing sequences arises in all three contexts, and supplemented by sieving methods, generating functions, asymptotics, and heuristic algorithms. In the area of number theory, this work generalizes the sieve of Eratosthenes to a sequence of polynomial values called polynomial-value sieving. In the case of quadratics, the method of polynomial-value sieving may be characterized briefly as a product presentation of two binary quadratic forms. Polynomials for which the polynomial-value sieving yields all possible integer factorizations of the polynomial values are called recursively-factorable. The Euler and Legendre prime producing polynomials of the form n2+n+p and 2n2+p, respectively, and Landau's n2+1 are shown to be recursively-factorable. Integer factorizations realized by the polynomial-value sieving method, applied to quadratic functions, are in direct correspondence with the lattice point solutions (X,Y) of the conic sections aX2+bXY +cY2+X-nY=0. The factorization structure of the underlying quadratic polynomial is shown to have geometric properties in the space of the associated lattice point solutions of these conic sections. In the area of combinatorial graph theory, this work considers two topological structures that are used to model the process of homologous genetic recombination: assembly graphs and chord diagrams. The result of a homologous recombination can be recorded as a sequence of signed permutations called a micronuclear arrangement. In the assembly graph model, each micronuclear arrangement corresponds to a directed Hamiltonian polygonal path within a directed assembly graph. Starting from a given assembly graph, we construct all the associated micronuclear arrangements. Another way of modeling genetic rearrangement is to represent precursor and product genes as a sequence of blocks which form arcs of a circle. Associating matching blocks in the precursor and product gene with chords produces a chord diagram. The braid index of a chord diagram can be used to measure the scope of interaction between the crossings of the chords. We augment the brute force algorithm for computing the braid index to utilize a divide and conquer strategy. Both assembly graphs and chord diagrams are closely associated with double occurrence words, so we classify and enumerate the double occurrence words based on several notions of irreducibility. In the area of analytic probability, moments abstractly describe the shape of a probability distribution. Over the years, numerous varieties of moments such as central moments, factorial moments, and cumulants have been developed to assist in statistical analysis. We use inversion formulas to compute high order moments of various types for common probability distributions, and show how the successive ratios of moments can be used for distribution and parameter fitting. We consider examples for both simulated binomial data and the probability distribution affiliated with the braid index counting sequence. Finally we consider a sequence of multiparameter binomial sums which shares similar properties with the moment sequences generated by the binomial and beta-binomial distributions. This sequence of sums behaves asymptotically like the high order moments of the beta distribution, and has completely monotonic properties.

This thesis presents the practical and novel work in the area of optimal placement of actuators and sensors for sound-structure interaction systems. The work has been done by the author during his PhD candidature. The research is concentrated in systems with non-ideal boundary conditions as in the case in practical engineering applications. An experimental acoustic cavity with five walls of timber and a thin aluminium sheet fixed tightly on the cavity mouth is chosen in this thesis as a good representation of general sound-structure interaction systems. The sheet is intentionally so fixed that it does not satisfy ideal boundary conditions. The existing methods for obtaining optimal sensor-actuator location using analytic models with ideal boundary conditions are of limited use for such problem with non-ideal boundary conditions. The method presented in this thesis for optimal placement of actuators and sensors is motivated by energy based approach and model uncertainty inclusion. The optimal placement of actuator and sensor for the experimental acoustic cavity is used to construct a robust feedback controller based on minimax LQG control design method. The controller is aimed to reduce acoustic potential energy in the cavity. This energy is due to the structure-borne sound inside the sound-structure interaction system. Practical aspects of the method for optimal placement of actuator and sensors are highlighted by experimental vibration and acoustic noise attenuation for arbitrary disturbance using feedback controllers with optimal placement of actuator and sensor. The disturbance is experimentally set to enter the system via a spatial location different from the controller input as would be in any practical applications of standard feedback disturbance rejections. Experimental demonstration of the novel methods presented in this thesis attenuate structural vibration up to 13 dB and acoustic noise up to 5 dB for broadband frequency range of interest. This attenuation is achieved without the explicit knowledge of the model of the disturbance.

<p>Algebraic equations and functions play an important role in various mathematical topics, including algebra, trigonometry, linear programming and calculus. Accordingly, various documents, such as the most recent Swedish curriculum (Lpf 94) for upper secondary school and the course syllabi in mathematics, specify what the students should learn in Mathematics Course B. They should be able to solve quadratic equations and apply this knowledge in solving problems, explain the properties of a function, as well as be able to set up, interpret and use some nonlinear functions as models for real processes. To implement these recommendations, it is crucial to understand the students’ way of experiencing quadratic equations and functions, and describe the meaning these have for the students in relation to the possibility they have to their experience of them.</p><p>The aim of this thesis is to analyse, understand and explain the relation between the handled and learned content, which consists of second-degree equations and quadratic functions, in classroom practice. This means that content is the research object and not the teacher’s conceptions or knowledge of, or about this content. This restriction implies that the handled and learned contents are central in this study and will be analysed from different perspectives.</p><p>The study includes two teachers and 45 students in two different classes. The data consist of video-recordings of lessons, individual sessions, interviews and the teachers’/researcher’s review of the individual sessions. The students’ tests also constituted an important part of the data collection.</p><p>When analysing the data, concepts relating to variation theory have been used as analytical tools. Data have been analysed in respect of the teachers’ focus on the lesson content, which aspects are ignored and which patterns of dimensions of variations are constituted when the contents are handled by the teachers in the classroom. Also, data have been analysed in respect of the students’ focus when they solve different exercises in a test situation. It can be shown that the meaning of parameters, the unknown quantity in an equation and the function’s argument change several times when the teacher presents the content in the classroom and when the students solve different exercises. It can also be shown that the teachers and the students develop complicated patterns of variation during the lessons and that the ways in which the teachers open up dimensions of variation play an important role in the learning process. The results indicate that there is a convergent variation leading the students to improve their learning. By focusing on some aspects of the objects of learning and create convergent variations, it is possible for the students to understand the difference between various interpretations of these aspects and thereafter focus on the interpretation that fits in a certain context. Furthermore, this variation leads the students to make generalisations in each object of learning (equations and functions) and between these objects of learning. These generalisations remain over time, despite working with new objects of learning. An important result in this study is that the implicit or explicit arguments of a function can make it possible to discern an equation from a function despite the fact that they are constituted by the same algebraic expression.</p>

<p>This thesis aims at implementing methods for factorisation of large numbers. Seeing that there is no deterministic algorithm for finding the prime factors of a given number, the task proves rather difficult. Luckily, there have been developed some effective probabilistic methods since the invention of the computer so that it is now possible to factor numbers having about 200 decimal digits. This however consumes a large amount of resources and therefore, virtually all new factorisations are achieved using the combined power of many computers in a distributed system. </p><p>The nature of the distributed system can vary. The original goal of the thesis was to develop a client/server system that allows clients to carry out a portion of the overall computations and submit the result to the server. </p><p>Methods for factorisation discussed for implementation in the thesis are: the quadratic sieve, the number field sieve and the elliptic curve method. Actually implemented was only a variant of the quadratic sieve: the multiple polynomial quadratic sieve (MPQS).</p>

The solution of a single optimization problem often requires computationally-demanding evaluations; this is especially true in optimal design of engineering components and systems described by partial differential equations. We present a technique for the rapid and reliable optimization of systems characterized by linear-functional outputs of partial differential equations with affine parameter dependence. The critical ingredients of the method are: (i) reduced-basis techniques for dimension reduction in computational requirements; (ii) an "off-line/on-line" computational decomposition for the rapid calculation of outputs of interest and respective sensitivities in the limit of many queries; (iii) a posteriori error bounds for rigorous uncertainty and feasibility control; (iv) Interior Point Methods (IPMs) for efficient solution of the optimization problem; and (v) a trust-region Sequential Quadratic Programming (SQP) interpretation of IPMs for treatment of possibly non-convex costs and constraints. / Singapore-MIT Alliance (SMA)

This thesis consists of four independent papers concerningdifferent aspects of interior methods for optimization. Threeof the papers focus on theoretical aspects while the fourth oneconcerns some computational experiments. The systems of equations solved within an interior methodapplied to a convex quadratic program can be viewed as weightedlinear least-squares problems. In the first paper, it is shownthat the sequence of solutions to such problems is uniformlybounded. Further, boundedness of the solution to weightedlinear least-squares problems for more general classes ofweight matrices than the one in the convex quadraticprogramming application are obtained as a byproduct. In many linesearch interior methods for nonconvex nonlinearprogramming, the iterates can "falsely" converge to theboundary of the region defined by the inequality constraints insuch a way that the search directions do not converge to zero,but the step lengths do. In the sec ond paper, it is shown thatthe multiplier search directions then diverge. Furthermore, thedirection of divergence is characterized in terms of thegradients of the equality constraints along with theasymptotically active inequality constraints. The third paper gives a modification of the analytic centerproblem for the set of optimal solutions in linear semidefiniteprogramming. Unlike the normal analytic center problem, thesolution of the modified problem is the limit point of thecentral path, without any strict complementarity assumption.For the strict complementarity case, the modified problem isshown to coincide with the normal analytic center problem,which is known to give a correct characterization of the limitpoint of the central path in that case. The final paper describes of some computational experimentsconcerning possibilities of reusing previous information whensolving system of equations arising in interior methods forlinear programming. <b>Keywords:</b>Interior method, primal-dual interior method,linear programming, quadratic programming, nonlinearprogramming, semidefinite programming, weighted least-squaresproblems, central path. <b>Mathematics Subject Classification (2000):</b>Primary90C51, 90C22, 65F20, 90C26, 90C05; Secondary 65K05, 90C20,90C25, 90C30.

Algebraic equations and functions play an important role in various mathematical topics, including algebra, trigonometry, linear programming and calculus. Accordingly, various documents, such as the most recent Swedish curriculum (Lpf 94) for upper secondary school and the course syllabi in mathematics, specify what the students should learn in Mathematics Course B. They should be able to solve quadratic equations and apply this knowledge in solving problems, explain the properties of a function, as well as be able to set up, interpret and use some nonlinear functions as models for real processes. To implement these recommendations, it is crucial to understand the students’ way of experiencing quadratic equations and functions, and describe the meaning these have for the students in relation to the possibility they have to their experience of them. The aim of this thesis is to analyse, understand and explain the relation between the handled and learned content, which consists of second-degree equations and quadratic functions, in classroom practice. This means that content is the research object and not the teacher’s conceptions or knowledge of, or about this content. This restriction implies that the handled and learned contents are central in this study and will be analysed from different perspectives. The study includes two teachers and 45 students in two different classes. The data consist of video-recordings of lessons, individual sessions, interviews and the teachers’/researcher’s review of the individual sessions. The students’ tests also constituted an important part of the data collection. When analysing the data, concepts relating to variation theory have been used as analytical tools. Data have been analysed in respect of the teachers’ focus on the lesson content, which aspects are ignored and which patterns of dimensions of variations are constituted when the contents are handled by the teachers in the classroom. Also, data have been analysed in respect of the students’ focus when they solve different exercises in a test situation. It can be shown that the meaning of parameters, the unknown quantity in an equation and the function’s argument change several times when the teacher presents the content in the classroom and when the students solve different exercises. It can also be shown that the teachers and the students develop complicated patterns of variation during the lessons and that the ways in which the teachers open up dimensions of variation play an important role in the learning process. The results indicate that there is a convergent variation leading the students to improve their learning. By focusing on some aspects of the objects of learning and create convergent variations, it is possible for the students to understand the difference between various interpretations of these aspects and thereafter focus on the interpretation that fits in a certain context. Furthermore, this variation leads the students to make generalisations in each object of learning (equations and functions) and between these objects of learning. These generalisations remain over time, despite working with new objects of learning. An important result in this study is that the implicit or explicit arguments of a function can make it possible to discern an equation from a function despite the fact that they are constituted by the same algebraic expression.

Neural networks play an important role in the execution of goal-oriented paradigms. They offer flexibility, adaptability and versatility, so that a variety of approaches may be used to meet a specific goal, depending upon the circumstances and the requirements of the design specifications. Development of higher-order neural units with higher-order synaptic operations will open a new window for some complex problems such as control of aerospace vehicles, pattern recognition, and image processing. The neural models described in this thesis consider the behavior of a single neuron as the basic computing unit in neural information processing operations. Each computing unit in the network is based on the concept of an idealized neuron in the central nervous system (CNS). Most recent mathematical models and their architectures for neuro-control systems have generated many theoretical and industrial interests. Recent advances in static and dynamic neural networks have created a profound impact in the field of neuro-control. Neural networks consisting of several layers of neurons, with linear synaptic operation, have been extensively used in different applications such as pattern recognition, system identification and control of complex systems such as flexible structures, and intelligent robotic systems. The conventional linear neural models are highly simplified models of the biological neuron. Using this model, many neural morphologies, usually referred to as multilayer feedforward neural networks (MFNNs), have been reported in the literature. The performance of the neurons is greatly affected when a layer of neurons are implemented for system identification, pattern recognition and control problems. Through simulation studies of the XOR logic it was concluded that the neurons with linear synaptic operation are limited to only linearly separable forms of pattern distribution. However, they perform a variety of complex mathematical operations when they are implemented in the form of a network structure. These networks suffer from various limitations such as computational efficiency and learning capabilities and moreover, these models ignore many salient features of the biological neurons such as time delays, cross and self correlations, and feedback paths which are otherwise very important in the neural activity. In this thesis an effort is made to develop new mathematical models of neurons that belong to the class of higher-order neural units (HONUs) with higher-order synaptic operations such as quadratic and cubic synaptic operations. The advantage of using this type of neural unit is associated with performance of the neurons but the performance comes at the cost of exponential increase in parameters that hinders the speed of the training process. In this context, a novel method of representation of weight parameters without sacrificing the neural performance has been introduced. A generalised representation of the higher-order synaptic operation for these neural structures was proposed. It was shown that many existing neural structures can be derived from this generalized representation of the higher-order synaptic operation. In the late 1960s, McCulloch and Pitts modeled the stimulation-response of the primitive neuron using the threshold logic. Since then, it has become a practice to implement the logic circuits using neural structures. In this research, realization of the logic circuits such as OR, AND, and XOR were implemented using the proposed neural structures. These neural structures were also implemented as neuro-controllers for the control problems such as satellite attitude control and model reference adaptive control. A comparative study of the performance of these neural structures compared to that of the conventional linear controllers has been presented. The simulation results obtained in this research were applicable only for the simplified model presented in the simulation studies.

Neural networks play an important role in the execution of goal-oriented paradigms. They offer flexibility, adaptability and versatility, so that a variety of approaches may be used to meet a specific goal, depending upon the circumstances and the requirements of the design specifications. Development of higher-order neural units with higher-order synaptic operations will open a new window for some complex problems such as control of aerospace vehicles, pattern recognition, and image processing. The neural models described in this thesis consider the behavior of a single neuron as the basic computing unit in neural information processing operations. Each computing unit in the network is based on the concept of an idealized neuron in the central nervous system (CNS). Most recent mathematical models and their architectures for neuro-control systems have generated many theoretical and industrial interests. Recent advances in static and dynamic neural networks have created a profound impact in the field of neuro-control. Neural networks consisting of several layers of neurons, with linear synaptic operation, have been extensively used in different applications such as pattern recognition, system identification and control of complex systems such as flexible structures, and intelligent robotic systems. The conventional linear neural models are highly simplified models of the biological neuron. Using this model, many neural morphologies, usually referred to as multilayer feedforward neural networks (MFNNs), have been reported in the literature. The performance of the neurons is greatly affected when a layer of neurons are implemented for system identification, pattern recognition and control problems. Through simulation studies of the XOR logic it was concluded that the neurons with linear synaptic operation are limited to only linearly separable forms of pattern distribution. However, they perform a variety of complex mathematical operations when they are implemented in the form of a network structure. These networks suffer from various limitations such as computational efficiency and learning capabilities and moreover, these models ignore many salient features of the biological neurons such as time delays, cross and self correlations, and feedback paths which are otherwise very important in the neural activity. In this thesis an effort is made to develop new mathematical models of neurons that belong to the class of higher-order neural units (HONUs) with higher-order synaptic operations such as quadratic and cubic synaptic operations. The advantage of using this type of neural unit is associated with performance of the neurons but the performance comes at the cost of exponential increase in parameters that hinders the speed of the training process. In this context, a novel method of representation of weight parameters without sacrificing the neural performance has been introduced. A generalised representation of the higher-order synaptic operation for these neural structures was proposed. It was shown that many existing neural structures can be derived from this generalized representation of the higher-order synaptic operation. In the late 1960s, McCulloch and Pitts modeled the stimulation-response of the primitive neuron using the threshold logic. Since then, it has become a practice to implement the logic circuits using neural structures. In this research, realization of the logic circuits such as OR, AND, and XOR were implemented using the proposed neural structures. These neural structures were also implemented as neuro-controllers for the control problems such as satellite attitude control and model reference adaptive control. A comparative study of the performance of these neural structures compared to that of the conventional linear controllers has been presented. The simulation results obtained in this research were applicable only for the simplified model presented in the simulation studies.

After introducing concepts from convex analysis, we study how to continuously transform one convex function into another. A natural choice is the arithmetic average, as it is pointwise continuous; however, this choice fails to average functions with different domains. On the contrary, the proximal average is not only continuous (in the epi-topology) but can actually average functions with disjoint domains. In fact, the proximal average not only inherits strict convexity (like the arithmetic average) but also inherits smoothness and differentiability (unlike the arithmetic average). Then we introduce a computational framework for computer-aided convex analysis. Motivated by the proximal average, we notice that the class of piecewise linear-quadratic (PLQ) functions is closed under (positive) scalar multiplication, addition, Fenchel conjugation, and Moreau envelope. As a result, the PLQ framework gives rise to linear-time and linear-space algorithms for convex PLQ functions. We extend this framework to nonconvex PLQ functions and present an explicit convex hull algorithm. Finally, we discuss a method to find primal-dual symmetric antiderivatives from cyclically monotone operators. As these antiderivatives depend on the minimal and maximal Rockafellar functions [5, Theorem 3.5, Corollary 3.10], it turns out that the minimal and maximal function in [12, p.132,p.136] are indeed the same functions. Algorithms used to compute these antiderivatives can be formulated as shortest path problems.