Variational inference for diffusion processes

Diffusion processes are a family of continuous-time continuous-state stochastic processes that are in general only partially observed. The joint estimation of the forcing parameters and the system noise (volatility) in these dynamical systems is a crucial, but non-trivial task, especially when the system is nonlinear and multimodal. We propose a variational treatment of diffusion processes, which allows us to compute type II maximum likelihood estimates of the parameters by simple gradient techniques and which is computationally less demanding than most MCMC approaches. We also show how a cheap estimate of the posterior over the parameters can be constructed based on the variational free energy.

Concrete beams with stirrups subject to torsion, shear and bending

This thesis examines experimentally and theoretically the behaviour and ultimate strength of rectangular reinforced concrete members under combined torsion, shear and bending. The experimental investigation consists of the test results of 38 longitudinally and transversely reinforced concrete beams subjected to combined loads, ten beams of which were tested under pure torsion and self-weight. The behaviour of each test beam from application of the first increment of load until failure is presented. The effects of concrete strength, spacing of the stirrups, the amount of longitudinal steel and the breadth of the section on the ultimate torsional capacity are investigated. Based on the skew-bending mechanism, compatibility, and linear stress-strain relationship for the concrete and the steel, simple rational equations are derived for the three principal modes of failure for the following four types of failure observed in the tests: TYPE I Yielding the reinforcement, at failure, before crushing the concrete. TYPE II Yielding of the web steel only, at failure, before crushing the concrete. TYPE III Yielding of the longitudinal steel only, at failure, before crushing the concrete. TYPE IV Crushing of the concrete, at failure, before yielding of any of the reinforcement.

Charge-balanced copolymer hydrogels:potential as biomedical materials

Polyzwitterionic-containing hydrogel materials been proposed for use in biomaterial applications. Polyzwitterions contain anions and cations in the same monomeric unit, unlike polyampholytes which contain them in different monomeric units. The use of cationic and anionic monomers in stoichiometrically equivalent proportions produces charge-balanced polyampholytes (PA) copolymers. Membranes prepared using either betaine-containing (BT) polyzwitterionic copolymers or PA copolymers can share similar properties, but the range of EWCs offered by membranes incorporating BT and PA monomers is greater than that for conventional neutral hydrogels and methacrylic acid-based systems. Here we compare properties of BT-containing and PA-containing copolymer membranes, relevant to their potential as biomedical materials. Membranes of the copolymers were prepared as previously described. Surface energy was determined using a GBX Digidrop (GBX Scientific Instruments), with diidomethane and water as probes. The absorption of proteins was determined by soaking the membranes in 1mg/ml protein solutions for a predetermined time, and measuring UV absorption of the membranes at certain wavelengths. The BT and PA copolymer membranes displayed similar values for the polar components and dispersive components of total surface free energy. This was perhaps not surprising when the structures of the monomers were considered. The BT and PA copolymer membranes displayed differences in their protein absorption over time, with the PA demonstrating higher uptake of protein than the BT. In addition to the aforementioned greater EWC range, the use of BT and PA copolymer membranes also avoids some of the problems associated with net anionicity. Comparison of the BT copolymer with the “pseudo” zwitterionic PA copolymers shows that controlled molecular architecture is required to gain the benefits of balancing the charges present in the copolymers in a way that will make them beneficial to hydrogel design.

Charge effects on the hindered transport of macromolecules across the endothelial surface glycocalyx layer

A fluid mechanical and electrostatic model for the transport of solute molecules across the vascular endothelial surface glycocalyx layer (EGL) was developed to study the charge effect on the diffusive and convective transport of the solutes. The solute was assumed to be a spherical particle with a constant surface charge density, and the EGL was represented as an array of periodically arranged circular cylinders of like charge, with a constant surface charge density. By combining the fluid mechanical analyses for the flow around a solute suspended in an electrolyte solution and the electrostatic analyses for the free energy of the interaction between the solute and cylinders based on a mean field theory, we estimated the transport coefficients of the solute across the EGL. Both of diffusive and convective transports are reduced compared to those for an uncharged system, due to the stronger exclusion of the solute that results from the repulsive electrostatic interaction. The model prediction for the reflection coefficient for serum albumin agreed well with experimental observations if the charge density in the EGL is ranged from approximately -10 to -30 mEq/l.

Applicability of contact angle techniques used in the analysis of contact lenses, part 1:comparative methodologies

Objectives and Methods: Contact angle, as a representative measure of surface wettability, is often employed to interpret contact lens surface properties. The literature is often contradictory and can lead to confusion. This literature review is part of a series regarding the analysis of hydrogel contact lenses using contact angle techniques. Here we present an overview of contact angle terminology, methodology, and analysis. Having discussed this background material, subsequent parts of the series will discuss the analysis of contact lens contact angles and evaluate differences in published laboratory results. Results: The concepts of contact angle, wettability and wetting are presented as an introduction. Contact angle hysteresis is outlined and highlights the advantages in using dynamic analytical techniques over static methods. The surface free energy of a material illustrates how contact angle analysis is capable of providing supplementary surface characterization. Although single values are able to distinguish individual material differences, surface free energy and dynamic methods provide an improved understanding of material behavior. The frequently used sessile drop, captive bubble, and Wilhelmy plate techniques are discussed. Their use as both dynamic and static methods, along with the advantages and disadvantages of each technique, is explained. Conclusions: No single contact angle technique fully characterizes the wettability of a material surface, and the application of complimenting methods allows increased characterization. At present, there is not an ISO standard method designed for soft materials. It is important that each contact angle technique has a standard protocol, as small protocol differences between laboratories often contribute to a variety of published data that are not easily comparable. © 2013 Contact Lens Association of Ophthalmologists.

Anisotropic surface chemistry of aspirin crystals

The wettability of the (001), (100), and (011) crystallographic facets of macroscopic aspirin crystals has been experimentally investigated using a sessile drop contact angle (θ) method. θ for a nonpolar liquid was very similar for all three facets, though significant θ differences were observed for three polar probe liquids. The observed hydrophobicity of the (001) and (100) facets is ascribed to a reduced hydrogen bonding potential at these surfaces, whilst the observed hydrophilicity of facet (011) may be attributed to presence of surface carboxylic functionalities as confirmed by X-ray photoelectron spectroscopy (XPS). The dispersive component of the surface free energy (γ) was similar for all three facets (35 ± 2 mJ/m). The total surface energy, γs varied between 46 and 60 mJ/m due to significant variations in the polar/acid-base components of γ for all facets. Surface polarity as determined by γ measurements and XPS data were in good agreement, linking the variations in wettability to the concentration of oxygen containing surface functional groups. In conclusion, the wettability and the surface energy of a crystalline organic solid, such as aspirin, was found to be anisotropic and facet dependant, and in this case, related to the presence of surface carboxylic functionalities. © 2007 Wiley-Liss, Inc. and the American Pharmacists Association.

An XPS study of pulsed plasma polymerised allyl alcohol film growth on polyurethane

The growth of highly functionalised poly allyl alcohol films by pulsed plasma polymerisation of CH2 double bond; length as m-dashCHCH2OH on biomedical grade polyurethane has been followed by X-ray photoelectron spectroscopy (XPS) and contact angle measurements. Film thickness is observed to increase approximately linearly with plasma modification time, suggesting a layer-by-layer growth mode of poly allyl alcohol. Water contact angle measurements reveal the change in the surface free energy of wetting decreases linearly with plasma modification up to the monolayer point after which a constant limiting value of −24 mJ m−2 was attained. Films prepared at 20 W plasma power with a duty cycle of 10 μs:500 μs exhibit a high degree of hydroxyl (single bondOH) retention with minimal fragmentation of the monomer observed. Increasing the plasma power up to 125 W is found to improve single bondOH retention at the expense of ether formation generating films close to the monomer stoichiometry. Duty cycle plays an important role in controlling both film composition and thickness, with longer off times increasing single bondOH retention, while longer on times enhance allyl alcohol film growth.

Variational mean-field algorithm for efficient inference in large systems of stochastic differential equations

This work introduces a Gaussian variational mean-field approximation for inference in dynamical systems which can be modeled by ordinary stochastic differential equations. This new approach allows one to express the variational free energy as a functional of the marginal moments of the approximating Gaussian process. A restriction of the moment equations to piecewise polynomial functions, over time, dramatically reduces the complexity of approximate inference for stochastic differential equation models and makes it comparable to that of discrete time hidden Markov models. The algorithm is demonstrated on state and parameter estimation for nonlinear problems with up to 1000 dimensional state vectors and compares the results empirically with various well-known inference methodologies.