This thesis, which consists of four chapters, focuses on three topics: discriminating between stationary and nonstationary time series, testing the constancy of the error covariance matrix of a vector model, and estimating density functions over bounded domains using kernel techniques. In Chapter 1, “Testing the unit root hypothesis against the logistic smooth transition autoregressive model”, and Chapter 2, “A nonlinear alternative to the unit root hypothesis”, the joint hypothesis of unit root and linearity allows one to distinguish between random walk processes, with or without drift, and stationary nonlinear processes of the smooth transition autoregressive type. This is important in applications because steps taken in modelling a time series are likely to be drastically different depending on whether or not the unit root hypothesis is rejected. In Chapter 1 the nonlinearity is based on the logistic function, while Chapter 2 considers the second-order logistic function. Monte Carlo simulations show that the proposed tests have about the same or higher power than the standard Dickey-Fuller unit root tests when the alternative exhibits nonlinear behavior. In Chapter 1 the tests are applied to the seasonally adjusted U.S. monthly unemployment rate, giving support to the hypothesis that the unemployment rate series follows a smooth transition autoregressive model rather than a random walk. Chapter 2 considers testing the so called purchasing power parity (PPP) hypothesis. The test results complement earlier studies, giving support to the PPP hypothesis for 44 out of 120 real exchange rates considered. Chapter 3. “Testing the constancy of the error covariance matrix in vector models”Estimating the parameters of an econometric model is necessary for any use of the model, be it forecasting or policy evaluation. Finding out thereafter whether or not the model appears to satisfy the assumptions under which it was estimated should be an integral part of a normal modelling exercise. This chapter includes the derivation of a Lagrange Multiplier test of the null hypothesis of constant variance in vector models when testing against three specific parametric alternatives. The Monte Carlo simulations show that the test has good size properties, very good power against a correctly specified alternative, but low or only up to moderate power in cases for a misspecified alternative hypothesis. Chapter 4. “ Estimating confidence regions over bounded domains”Nonparametric density estimation by kernel techniques is a standard statistical tool in the estimation of a density function in situations where its parametric form is assumed to be unknown. In many cases, the data set over which the density is to be estimated exhibits linear, or nonlinear, dependence. A solution to this problem is to apply a one-to-one transformation to the considered data set in such a way that the dependence in the data vanishes, but too often such a unique transformation does not exist. This chapter proposes a method for estimating confidence regions over bounded domains when no one-to-one transformation of the considered data exists, or if the existence of such a transformation is difficult to verify. The method, simple kernel estimation over a nonlinear grid, is illustrated by applying it to three data sets generated from the GARCH(1,1) model. The resulting confidence regions cover a reasonable area of the definition space, and are well aligned with the corresponding data sets. / Diss. Stockholm : Handelshögsk., 2003

Boolean functions play an important role in the design and analysis of symmetric-key cryptosystems, as well as having applications in other fields such as coding theory. Boolean functions acting on large number of inputs introduces the problem of computing the cryptographic properties. Traditional methods of computing these properties involve transformations which require computation and memory resources exponential in the number of input variables. When the number of inputs is large, Boolean functions are usually defined by the algebraic normal form (ANF) representation. In this thesis, methods for computing the weight and nonlinearity of Boolean functions from the ANF representation are investigated. The relation between the ANF coecients and the weight of a Boolean function was introduced by Carlet and Guillot. This expression allows the weight to be computed in $mathcal{O}(2^p)$ operations for a Boolean function containing p monomials in its ANF. In this work, a more ecient algorithm for computing the weight is proposed, which eliminates the unnecessary calculations in the weight expression. By generalizing the weight expression, a formulation of the distances to the set of linear functions is obtained. Using this formulation, the problem of computing the nonlinearity of a Boolean function from its ANF is reduced to an associated binary integer programming problem. This approach allows the computation of nonlinearity for Boolean functions with high number of input variables and consisting of small number of monomials in a reasonable time.

Boolean functions play an important role in the design and analysis of symmetric-key cryptosystems, as well as having applications in other fields such as coding theory. Boolean functions acting on large number of inputs introduces the problem of computing the cryptographic properties. Traditional methods of computing these properties involve transformations which require computation and memory resources exponential in the number of input variables. When the number of inputs is large, Boolean functions are usually defined by the algebraic normal form (ANF) representation. In this thesis, methods for computing the weight and nonlinearity of Boolean functions from the ANF representation are investigated. The relation between the ANF coefficients and the weight of a Boolean function was introduced by Carlet and Guillot. This expression allows the weight to be computed in $mathcal{O}(2^p)$ operations for a Boolean function containing $p$ monomials in its ANF. In this work, a more efficient algorithm for computing the weight is proposed, which eliminates the unnecessary calculations in the weight expression. By generalizing the weight expression, a formulation of the distances to the set of linear functions is obtained. Using this formulation, the problem of computing the nonlinearity of a Boolean function from its ANF is reduced to an associated binary integer programming problem. This approach allows the computation of nonlinearity for Boolean functions with high number of input variables and consisting of small number of monomials in a reasonable time.

The money demand function has been regarded as a fundamental building block in macroeconomic modelling, as it represents the link between the monetary policy and rest of the economy. The extensive literature on money demand function is concerned with the existence of a stable money demand function, which ensures adequate prediction of impact of a given change in money supply on other economic variables such as, inflation, interest rates, national income, private investment and other policy variables. This thesis employs both linear and nonlinear estimation methods to investigate the relationship between money demand, GDP, inflation and interest rates for the Euro Area over the period 1980-2010. The aim of this thesis is to compare the European money demand in linear and nonlinear framework. First a vector autoregression (VAR) model has been estimated. Then a threshold cointegration model has been employed and nonlinearity properties of the money demand relationship has been investigated. In contrast to the existing empirical literature, linear VEC model can find evidence of stability, however it has some conflicting results which can be explained by the nonlinearity of the model. Empirical results of MTAR type threshold cointegration specification verifies the nonlinearity in European money demand. The adjustment coefficient of lower regime suggests faster adjustment towards long run equilibrium compared to upper regime in nonlinear model. Moreover, the nonlinear model presents better fit to economic literature than linear model for European money demand.

Heterogeneity in lattice potentials (like random or quasiperiodic) can localize linear, non-interacting waves and halt their propagation. Nonlinearity induces wave interactions, enabling energy exchange and leading to chaotic dynamics. Understanding the interplay between the two is one of the topical problems of modern wave physics. In particular, one questions whether nonlinearity destroys localization and revives wave propagation, whether thresholds in wave energy/norm exist, and what the resulting wave transport mechanisms and characteristics are. Despite remarkable progress in the field, the answers to these questions remain controversial and no general agreement is currently achieved. This thesis aims at resolving some of the controversies in the framework of nonlinear dynamics and computational physics. Following common practice, basic lattice models (discrete Klein-Gordon and nonlinear Schroedinger equations) were chosen to study the problem analytically and numerically. In the disordered linear case all eigenstates of such lattices are spatially localized manifesting Anderson localization, while nonlinearity couples them, enabling energy exchange and chaotic dynamics. For the first time we present a comprehensive picture of different subdiffusive spreading regimes and self-trapping phenomena, explain the underlying mechanisms and derive precise asymptotics of spreading. Moreover, we have successfully generalized the theory to models with spatially inhomogeneous nonlinearity, quasiperiodic potentials, higher lattice dimensions and arbitrary nonlinearity index. Furthermore, we have revealed a remarkable similarity to the evolution of wave packets in the nonlinear diffusion equation. Finally, we have studied the limits of strong disorder and small nonlinearities to discover the probabilistic nature of Anderson localization in nonlinear disordered systems, demonstrating the finite probability of its destruction for arbitrarily small nonlinearity and exponentially small probability of its survival above a certain threshold in energy. Our findings give a new dimension to the theory of wave packet spreading in localizing environments, explain existing experimental results on matter and light waves dynamics in disordered and quasiperiodic lattice potentials, and offer experimentally testable predictions.

The resistor ladder (R/2R) digital-to-analogue converter (DAC) architecture is often used in high performance audio solutions due to its low-noise performance. Even high-end R/2R DACs suffer from static nonlinearity distortions. It was suspected that compensating for these nonlinearities would be possible. It was also suspected that this could improve audio quality in audio systems using R/2R DACs for digital-to-analogue (A/D) conversion. Through the use of models of the resistor ladder architecture a way of characterizing and measuring the faults in the R/2R DAC was created. A compensation algorithm was developed in order to compensate for the nonlinearities. The performance of the algorithm was simulated and an implementation of it was evaluated using an audio evaluation instrument. The results presented show that it is possible to increase linearity in R/2R DACs by compensating for static nonlinearity distortions. The increase in linearity can be quite significant and audible for the trained ear.

In this thesis, we analyze the performance of a loop-back built-in-self-test for a RF transceiver front-end. The tests aim at spot defects in a transceiver front-end and they make use of RF specifications such as NF (Noise Figure), G (power gain) and IIP3 (third order Intercept point). To enhance fault detectability, RF signal path sensitization is introduced. We use a functional RF transceiver model that is implemented in MatLab™ to verify this analysis.

Despite a large demand from industry, a realistic chatter modeling for hard turning has not been available due to the complexity of the problem, which is mainly caused by flank wear and nonlinearity in hard turning. This thesis attempts to develop chatter models for predicting chatter stability conditions in hard turning with the considerations of the effects of flank wear and nonlinearity. First, a linear model is developed by introducing non-uniform load distribution on a tool tip to account for the flank wear effect. Second, a nonlinear model is developed by further incorporating nonlinearity in the structure and cutting force. Third, stability analysis based on the root locus method and the describing function approach is conducted to determine a critical stability parameter. Fourth, to validate the models, a series of experiment is carried out to determine the stability limits as well as certain characteristic parameters for facing and straight turning. From these, it is shown that the nonlinear model provides more accurate predictions than the linear model, especially in the high-speed range. Furthermore, the stabilizing effect due to flank wear is confirmed through a series of experiments. Fifth, to fully account for the validity of linear and nonlinear models, an empirical model is proposed to fit in with the experimental stability limits in the full range of cutting speed. The proposed linear and nonlinear chatter models will help to improve the productivity in many manufacturing processes. In addition, chatter experimental data will be useful to develop other chatter models in hard turning.

Utilizing the universal approximation property of neural networks, we develop several novel approaches to neural network-based adaptive output feedback control of nonlinear systems, and illustrate these approaches for several flight control applications. In particular, we address the problem of non-affine systems and eliminate the fixed point assumption present in earlier work. All of the stability proofs are carried out in a form that eliminates an algebraic loop in the neural network implementation. An approximate input/output feedback linearizing controller is augmented with a neural network using input/output sequences of the uncertain system. These approaches permit adaptation to both parametric uncertainty and unmodeled dynamics. All physical systems also have control position and rate limits, which may either deteriorate performance or cause instability for a sufficiently high control bandwidth. Here we apply a method for protecting an adaptive process from the effects of input saturation and time delays, known as ``pseudo control hedging". This method was originally developed for the state feedback case, and we provide a stability analysis that extends its domain of applicability to the case of output feedback. The approach is illustrated by the design of a pitch-attitude flight control system for a linearized model of an R-50 experimental helicopter, and by the design of a pitch-rate control system for a 58-state model of a flexible aircraft consisting of rigid body dynamics coupled with actuator and flexible modes. A new approach to augmentation of an existing linear controller is introduced. It is especially useful when there is limited information concerning the plant model, and the existing controller. The approach is applied to the design of an adaptive autopilot for a guided munition. Design of a neural network adaptive control that ensures asymptotically stable tracking performance is also addressed.

Thermally cycling a thin plate of nickel-based superalloy with an intense in-plane thermal gradient, or hot spot, produces thermally induced crack growth not represented by classic thermo-mechanical fatigue (TMF). With the max hot spot temperature at 1093 C (2000 F) of a 1.5 mm thick, 82.55 mm diameter circular plate of B-1900+Hf, annular buckling and bending stresses result during each thermal cycle which drive the crack initiation and propagation. A finite element analysis (FEA) model, using ANSYS 7.1, has been developed which models the buckling and as well as represents the stress intensity at simulated crack lengths upon cool down of each thermal cycle. The model approximates the out-of-plane response at heat-up within 5% error and a difference in the final displacement of 0.185 mm after twelve thermal cycles. Using published da/dN vs. Keff data, the number of cycles needed to grow the crack to the experimental arrest distance is modeled within 1 mm. The number of cycles to this point is within 5 out of 462 in comparison to the experimental test.