The performance of modern VLSI systems is limited by signal transmission characteristics of electrical interconnections. Free-space optical interconnects have been suggested as a method to solve these problems. In this work, holographic optical elements (HOEs) for use as free-space interconnects and optical receivers compatible with CMOS systems are investigated. First, the switching characteristics of two basic CMOS receivers without a gain stage are investigated. A simple low-impedance load configuration for high-speed operation is introduced which requires only one photodetector to receive optical signals, and one transistor to provide a desired bias. The optimization and various operating characteristics of this receiver are discussed in detail. SPICE simulations and experimental results using discrete components demonstrate that this receiver can operate at high-frequencies with reasonable optical input power. However, the system fan-out is limited by the available optical power. In order to improve system fan-out, the basic low-impedance load CMOS receiver is extended with a simple gain stage without significantly increasing the complexity of the receivers. Addition of the gain stage is shown to improve fan-out by one order of magnitude at a fixed operating frequency. Free-space interconnects using HOEs have many advantages such as combining several optical functions into a single thin film element. However, the realization of these unique features is limited by image degradation effects due to misalignment and wavelength variations of the light source. Substrate-mode holograms (SMHs) are used to minimize these image degradation effects. Methods for recording SMHs in dichromated gelatin (DCG) are described. Techniques for fabricating SMHs and controlling the angular bandwidth are presented. Experimental results for different combinations of DCG SMHs show that these high-performance elements can function as free-space optical interconnects. In addition, highly polarization-selective substrate-mode holograms were investigated and fabricated. A demonstration of beam switching using polarization selective and non-selective elements with an electro-optic halfwave plate is also given.

We investigate stochastic systems which have a set of control parameters and a performance criterion. By operating the system at fixed control parameters, noisy performance values are observed. (The values are noisy due to the inherent stochastic nature of the system.) Certain relevant distributions of the system are assumed unavailable. The task is to develop algorithms that guide the system to optimal parameter settings based on its operating history. Consider the stochastic shortest path problem, where the time to traverse an edge is given by a random variable whose distribution is unavailable explicitly. The optimality criterion (to be maximized) is the probability of going from a given source node to a given terminal node within a specified critical time period. By choosing a particular path and traversing it, realizations of the distributions, i.e. time to go from the source node to the terminal node on that path are observed. Or consider the M/M/1 queue. Here, the control parameter is the average service time. The performance criterion is the sum of cost of the server and cost of system time of a customer in steady-state. By choosing a particular average service time and serving accordingly, a noisy observation of the total cost is obtained. We combine an asymptotically optimal random search method for finding the global optimum of a function with problem-specific local search techniques. Such a combination results in efficient solution procedures for the above problems. This conclusion is reached by applying the procedure to problems for which the optimum solutions are known. The main contribution of the study is in demonstrating that "reasonable" performance can be achieved for the proposed optimization problems in "reasonable" time by exploiting problem-specific structures to advantage. The generality of the method should allow others to use it in different optimization settings than ours. Also, the self-optimizing aspect of these methods and the stochastic versions of the local search techniques are new.

This dissertation is aimed at a class of convex dynamic optimization problems in which the transition functions are twice continuously differentiable and the stagewise objective functions are convex, although not necessarily differentiable. Two basic descent algorithms which use sequential and parallel coordinating techniques are developed. In both algorithms the nondifferentiability of the objective function is accounted for by using subgradient information. The objective of the subproblems generated consists of successive piecewise linear approximations of the stagewise objective function and the value function. In the parallel algorithm, an incentive coordination method is used to coordinate the subproblems. We provide proofs of convergence for these algorithms. Two variations, namely, subgradient selection and subgradient aggregation, of the basic algorithms are also discussed. In practice while subgradient selection seems to perform well, computational results with subgradient aggregation are rather disappointing. Computational results of the basic algorithms and variants based on subgradient selection are given. The effect of number of stages on performance of these algorithms is compared with a general nonlinear programming package (NPSOL).

A backpropagation neural network was trained to estimate the spatial location (offset and depth) of a target given an image of the electromagnetic ellipticity. A length of galvanized pipe buried 2m deep was used as a target. Three components of the magnetic field were measured from which the ellipticity was calculated. Finite element models of the target at different offsets and depths were used to calculate the theoretical ellipticity. The theoretical results were used for training the neural network. The network was tested on additional theoretical models and the field data. The input data representation is important in obtaining good results from the neural network; generally, the smaller the input vectors, the better the results. Five different representations were examined: the whole image, the subsampled image, trough-peak-trough, peak amplitude and frequency-domain. The frequency-domain representation estimated the target locations with the least error. The network was examined for its ability to generalize, to extrapolate beyond the spatial limits of the training set, and to ignore noise. The ability to generalize from theoretical training data to theoretical test data was good for all data representations. Extrapolation errors were satisfactory up to 1.5 model spacings away from the limits of the training set. The ability to ignore noise was generally best for smaller representations with the least amount of training. A third parameter, conductivity-area product was added to the network to more closely simulate the results from standard inversion routines and to test the ability of the network to scale to larger problems. The addition of multiple training examples for each model location improved the results. The increase in training set size dominated the scaling results. The time required for convergence increased exponentially with training set size. Data representation did not have as great an effect on training time.

A theoretical and experimental study of the phase behavior of aqueous salt-polymer two-phase systems has been done. A statistical mechanical model has been developed for the chemical potential of every component in the salt-polymer-water system. The model incorporates the effect of short-range forces by use of the isothermal-isobaric osmotic pressure expansion of Hill. The effect of long-range forces such as electrostatic interactions is incorporated with a non-primitive electrolyte model based on the work of Pailthrope et al. and on Kirkwood-Buff theory. The effect of polymer-polymer and polymer-salt interactions is represented in the model by Hill osmotic virial coefficients. The polymer molecular weight dependence of the second virial coefficients is predicted with the results of polymer scaling laws. An isopiestic experiment has been developed to measure the thermodynamic activity data required to evaluate the model parameters. Six different aqueous mixtures of polyethylene glycol 1000 and 8000 and MgSO₄, Na₂SO₄, and Na₂CO₃ were studied at 25°C, 1 ATM. Phase diagrams for these six systems were calculated from the model and compared to experiment with good agreement between them.

A new and powerful concept of modeling--the disturbed state concept is applied to the joint case. The disturbed state modeling is based on the assumption that the behavior of the joint, or the behavior at the disturbed state can be expressed by the joint behaviors at its reference states. The reference states include the original state and the critical state. The behavior of the intact joint at the original state is modeled by using a general plasticity joint model developed by Desai and Fishman (1987) with a small modification. The critical state joint is modeled according to observations from shear tests of the joints. The disturbed state joint model developed is capable of describing the hardening and softening behavior of the joint under various stress paths. Verification is made by back predictions and predictions of several series of shear tests. The constants used for back predictions are obtained from those tests back-predicted and the constants used for predictions are from the tests other than the tests predicted. Other important issues such as the size effect of joint sample sizes, the relationship between the roughness of the joint and the parameters in the joint model are also examined.

A constitutive model based on elasto-plasticity theory is proposed here to describe the behavior of interfaces subjected to static and cyclic loading conditions. The proposed model is developed in a hierarchical manner wherein a basic model describing simplified characteristics of the interfaces is modified by introducing different features, to model increasingly complex behavior of the interfaces. The proposed model can simulate associative, nonassociative, and strain-softening behavior during monotonic as well as cyclic loading. The parameters influencing interface behavior are identified using data from laboratory simple shear tests on sand-steel and sand-concrete interfaces. A parameter called "interface roughness ratio, R" is defined in order to model the interface behavior under different interface roughnesses. Similarly, a cyclic parameter Ω is introduced to simulate the cyclic volumetric behavior of the interfaces. Proposed model is verified with respect to comprehensive test data on interfaces with different roughnesses, normal loads, initial densities and type of sand, and quasi-static and cyclic loading. A new and highly efficient algorithm is developed to perform drift correction under constraint condition. This algorithm is used for the integration of constitutive relation for interfaces to perform back prediction. Performance of the algorithm is compared with various existing algorithms. Using Lyapunov's Stability Theorem, it is proved that the proposed algorithm is stable. The proposed model for the interfaces is used in the context of the thin-layer element approach and is implemented in a nonlinear dynamic finite element code to solve a boundary value problem involving dynamics of an axially loaded pile. It is shown here that the use of the interface model can allow proper modeling of shear transfer, volumetric behavior and localized relative slip in the interface zone. The effect on shear transfer from pile to soil due to the coupling between normal behavior and shear behavior of interface is established here for soil-structure interaction problems. The findings of this research have contributed to the understanding of the interface behavior in soil-structure interaction problems. The proposed model can simulate a number of important behavioral aspects of the interfaces.

Rock fragmentation can occur by crushing, relative radial motion, release of load, spalling, gas extension of strain wave-induced and pre-existing fractures, shear and in-flight collisions. The relative importance of each mechanism, as yet undefined, depends upon explosive properties, rock properties, blast geometry and initiation sequence. The effect of explosive energy and presence of natural fractures on fragmentation are studied. Design of blasts in a copper porphyry ore body were altered to give powder factors between 0.26 and 0.71 kg/m³ in three different types of rock. The rock units in the ore body for this study are classified according to the size distribution of the blocks formed by natural fractures and discontinuities. Block size distribution curves are produced from the core logs and the ore body is divided into six rock classes. Arbitrarily, a representative size distribution curve is assigned to each class. These curves are used to determine the specific surface area (surface area per unit volume of rock) for each rock class. The specific surface area of blasted rock was measured by photo-analysis and correlated closely with the explosive energy and rock type. The explosive energy is calculated in terms of total energy per unit volume of blasted rock, from the powder factor. The results clearly show the effect of natural block sizes on fragmentation. It is shown that most of useful explosive energy goes into opening the pre-existing fractures. Almost the entire product size is controlled by the natural fractures at energy levels below 2 MJ/m³ (powder factor of 0.5 kg/m³). At higher energy levels of 2.85 MJ/m³ (powder factor of 0.71 kg/m³) about 70 to 80 percent of the surface area of the blasted rock is controlled by the pre-existing fractures. The blasted rock is also analyzed using the scale invariant methods as an alternative to the conventional size distribution analysis. The fractal dimension of each blast at different energy level and rock class is determined. The variations of fractal dimensions depends on the energy levels and the rock conditions. At higher energy levels higher fractal dimensions are obtained and more fractured rock showed higher fractal dimension at the same energy level. The fractal concept can be a useful tool to describe rock masses and fragmenting behavior of different rocks.

A theoretical model for scanned focussed ultrasound hyperthermia was evaluated in canine muscle in vivo. This model is composed of two models: an ultrasonic power deposition model and a heat transfer model. One ultrasound model and two bio-heat transfer models were considered. (1) Ultrasound field distributions were measured using thermal techniques in both canine thighs in vivo and in water. The experimental results were compared with distributions obtained from a model based on the one dimensional integration of the Rayleigh-Sommerfeld diffraction integral. The comparisons showed that the model is a good approximation to the distributions measured in water. The main lobe profiles obtained in the muscle also agreed well with both model predictions and results measured in water. However, these in vivo distributions showed enlargement of the side lobes. It was also found that muscle interfaces produced considerable beam distortions and increased side lobes. These findings were verified by measurements of the peak intensity and the total acoustic power attenuation coefficients for passage of the beams through thighs that showed that the former was about 40% higher than the latter. Also, absolute intensities at the acoustic focus were measured in water with a hydrophone for 11 transducers ranging in frequency from 0.246 to 3.54 MHz. When these intensities were compared to model predictions, it was found that the model overestimated the peak intensity by a factor of less than 2. That is, the model can be used to obtain upper bounds for absolute intensity. (2) Steady state temperature profiles from a simple (uniform blood perfusion) three dimensional bio-heat transfer model (Pennes), and from a simple (isotropic thermal conductivity) three dimensional effective thermal conductivity model, were compared with temperatures measured during scanned focussed ultrasound hyperthermia experiments in canine thighs in vivo. The experimental data consisted of radial temperature profiles across single octagonal scans measured at different depths into the thighs. The results showed that the bio-heat transfer equation predicted the experimental trends qualitatively and that the effective thermal conductivity equation failed to do so. Both models failed to predict the influence of thermally significant vessels. A scanned focussed ultrasound model composed of the ultrasound model evaluated here and the bio-heat transfer equation, can be used to predict the major features of temperature fields for hyperthermia patient treatment planning.

A geometrical model for the rock crushed zone around a cylindrical charge is developed. The model is used to obtain empirical relationships between the scaled crushed zone diameter and some dimensionless ratios of explosive and rock properties. The ratios are velocity ratio, characteristic impedance ratio, medium stress ratio, and detonation pressure ratio. The empirical relations for granite, salt, and limestone in combination with a variety of explosives show that the scaled crushed zone diameter increases at a decreasing rate with increasing dimensionless ratios. The shape of the wave fronts around a cylindrical charge detonating in rock has been constructed for velocity ratios ranging from infinity to less than one. The shape of the wave front is not planar in the range of dimensions used in full scale bench blasting. The shape of the wave front is cylindrical in the middle and spherical at the top and bottom for infinite velocity ratio; sphero-conical for velocity ratios greater than one; spherical for velocity ratios ≤ 1. Quasi-static finite element models for a blasthole in a full scale bench blasting are analyzed using a 2-D finite element program written by the author. The models include a model neglecting radial cracks, models considering pressurized and non-pressurized radial cracks around the blasthole, and a model using an equivalent cavity to replace the pressurized radial cracks. Displacement fields, stress fields, and strain energy density distribution are studied. The analyses show that including radial cracks increases the levels of the strain energy density contours and the magnitudes of the displacement and stress fields several fold. The equivalent cavity gives much lower levels of strain energy contours and gives lower displacement and stress field magnitudes than those produced by the pressurized radial cracks. The scaled areas of the strain energy density contours increase at a decreasing rate with increasing the blasthole internal pressure and with increasing the ratio of the compressive strength to the tensile strength. These contour areas decrease at a decreasing rate with increasing tensile strength.