A basis function approach to Bayesian inference in diffusion processes

In this paper, we present a framework for Bayesian inference in continuous-time diffusion processes. The new method is directly related to the recently proposed variational Gaussian Process approximation (VGPA) approach to Bayesian smoothing of partially observed diffusions. By adopting a basis function expansion (BF-VGPA), both the time-dependent control parameters of the approximate GP process and its moment equations are projected onto a lower-dimensional subspace. This allows us both to reduce the computational complexity and to eliminate the time discretisation used in the previous algorithm. The new algorithm is tested on an Ornstein-Uhlenbeck process. Our preliminary results show that BF-VGPA algorithm provides a reasonably accurate state estimation using a small number of basis functions.

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.

Ionic and molecular diffusion in cementitious materials

The work described in this thesis is an attempt to provide improved understanding of the effects of several factors affecting diffusion in hydrated cement pastes and to aid the prediction of ionic diffusion processes in cement-based materials. Effect of pore structure on diffusion was examined by means of comparative diffusion studies of quaternary ammonium ions with different ionic radii. Diffusivities of these ions in hydrated pastes of ordinary portland cement with or without addition of fly ash were determined by a quasi-steady state technique. The restriction of the pore geometry on diffusion was evaluated from the change of diffusivity in response to the change of ionic radius. The pastes were prepared at three water-cement ratios, 0.35, 0.50 and 0.65. Attempts were made to study the effect of surface charge or the electrochemical double layer at the pore/solution interface on ionic diffusion. An approach was to evaluate the zeta potentials of hydrated cement pastes through streaming potential measurements. Another approach was the comparative studies of the diffusion kinetics of chloride and dissolved oxygen in hydrated pastes of ordinary portland cement with addition of 0 and 20% fly ash. An electrochemical technique for the determination of oxygen diffusivity was also developed. Non-steady state diffusion of sodium potassium, chloride and hydroxyl ions in hydrated ordinary portland cement paste of water-cement ratio 0.5 was studied with the aid of computer-modelling. The kinetics of both diffusion and ionic binding were considered for the characterization of the concentration profiles by Fick's first and second laws. The effect of the electrostatic interactions between ions on the overall diffusion rates was also considered. A general model concerning the prediction of ionic diffusion processes in cement-based materials has been proposed.

Opinion leaders' role in innovation diffusion:a simulation study

This study investigates the critical role that opinion leaders (or influentials) play in the adoption process of new products. Recent existing reseach evidence indicates a limited effect of opinion leaders on diffusion processes, yet these studies take into account merely the network position of opinion leaders without addressing their influential power. Empirical findings of our study show that opinion leaders, in addition to having a more central network position, possess more accurate knowledge about a product and tend to be less susceptible to norms and more innovative. Experiments that address these attributes, using an agent-based model, demonstrate that opinion leaders increase the speed of the information stream and the adoption process itself. Furthermore, they increase the maximum adoption percentage. These results indicate that targeting opinion leaders remains a valuable marketing strategy.

Derivations of variational gaussian process approximation framework

Recently, within the VISDEM project (EPSRC funded EP/C005848/1), a novel variational approximation framework has been developed for inference in partially observed, continuous space-time, diffusion processes. In this technical report all the derivations of the variational framework, from the initial work, are provided in detail to help the reader better understand the framework and its assumptions.

Variational Markov Chain Monte Carlo for Bayesian smoothing of non-linear niffusions

In this paper we develop set of novel Markov chain Monte Carlo algorithms for Bayesian smoothing of partially observed non-linear diffusion processes. The sampling algorithms developed herein use a deterministic approximation to the posterior distribution over paths as the proposal distribution for a mixture of an independence and a random walk sampler. The approximating distribution is sampled by simulating an optimized time-dependent linear diffusion process derived from the recently developed variational Gaussian process approximation method. Flexible blocking strategies are introduced to further improve mixing, and thus the efficiency, of the sampling algorithms. The algorithms are tested on two diffusion processes: one with double-well potential drift and another with SINE drift. The new algorithm's accuracy and efficiency is compared with state-of-the-art hybrid Monte Carlo based path sampling. It is shown that in practical, finite sample, applications the algorithm is accurate except in the presence of large observation errors and low observation densities, which lead to a multi-modal structure in the posterior distribution over paths. More importantly, the variational approximation assisted sampling algorithm outperforms hybrid Monte Carlo in terms of computational efficiency, except when the diffusion process is densely observed with small errors in which case both algorithms are equally efficient.

A new variational radial basis function approximation for inference in multivariate diffusions

In this paper we present a radial basis function based extension to a recently proposed variational algorithm for approximate inference for diffusion processes. Inference, for state and in particular (hyper-) parameters, in diffusion processes is a challenging and crucial task. We show that the new radial basis function approximation based algorithm converges to the original algorithm and has beneficial characteristics when estimating (hyper-)parameters. We validate our new approach on a nonlinear double well potential dynamical system.

Wear and corrosion of mechanical seals

Mechanical seals are used extensively to seal machinery such as pumps, mixers and agitators in the oil, petrochemical and chemical industries. The performance of such machinery is critically dependent on these devices. Seal failures may result in the escape of dangerous chemicals, possibly causing injury or loss of life. Seal performance is limited by the choice of face materials available. These range from cast iron and stellited stainless steel to cemented and silicon carbides. The main factors that affect seal performance are the wear and corrosion of seal faces. This research investigated the feasibility of applying surface coating/treatments to seal materials, in order to provide improved seal performance. Various surface coating/treatment methods were considered; these included electroless nickel plating, ion plating, plasma nitriding, thermal spraying and high temperature diffusion processes. The best wear resistance, as evaluated by the Pin-on-Disc wear test method, was conferred by the sprayed tungsten carbide/nickel/tungsten-chromium carbide deposit, produced by the high energy plasma spraying (Jet-Kote) process. In general, no correlation was found between hardness and wear resistance or surface finish and friction. This is due primarily to the complexity of the wear and frictional oxidation, plastic deformation, ploughing, fracture and delamination. Corrosion resistance was evaluated by Tafel extrapolation, linear polarisation and anodic potentiodynamic polarisation techniques. The best corrosion performance was exhibited by an electroless nickel/titanium nitride duplex coating due to the passivity of the titanium nitride layer in the acidified salt solution. The surface coating/treatments were ranked using a systematic method, which also considered other properties such as adhesion, internal stress and resistance to thermal cracking. The sealing behaviour of surface coated/treated seals was investigated on an industrial seal testing rig. The best sealing performances were exhibited by the Jet-Kote and electroless nickel silicon carbide composite coated seals. The failure of the electroless nickel and electroless nickel/titanium nitride duplex coated seals was due to inadequate adhesion of the deposits to the substrate. Abrasion of the seal faces was the principal wear mechanism. For operation in an environment similar to the experimental system employed (acidified salt solution) the Jet-Kote deposit appears to be the best compromise.

Networking - a statistical physics perspective

Networking encompasses a variety of tasks related to the communication of information on networks; it has a substantial economic and societal impact on a broad range of areas including transportation systems, wired and wireless communications and a range of Internet applications. As transportation and communication networks become increasingly more complex, the ever increasing demand for congestion control, higher traffic capacity, quality of service, robustness and reduced energy consumption requires new tools and methods to meet these conflicting requirements. The new methodology should serve for gaining better understanding of the properties of networking systems at the macroscopic level, as well as for the development of new principled optimization and management algorithms at the microscopic level. Methods of statistical physics seem best placed to provide new approaches as they have been developed specifically to deal with nonlinear large-scale systems. This review aims at presenting an overview of tools and methods that have been developed within the statistical physics community and that can be readily applied to address the emerging problems in networking. These include diffusion processes, methods from disordered systems and polymer physics, probabilistic inference, which have direct relevance to network routing, file and frequency distribution, the exploration of network structures and vulnerability, and various other practical networking applications. © 2013 IOP Publishing Ltd.

Variational Markov chain Monte Carlo for Bayesian smoothing of non-linear diffusions

In this paper we develop set of novel Markov Chain Monte Carlo algorithms for Bayesian smoothing of partially observed non-linear diffusion processes. The sampling algorithms developed herein use a deterministic approximation to the posterior distribution over paths as the proposal distribution for a mixture of an independence and a random walk sampler. The approximating distribution is sampled by simulating an optimized time-dependent linear diffusion process derived from the recently developed variational Gaussian process approximation method. The novel diffusion bridge proposal derived from the variational approximation allows the use of a flexible blocking strategy that further improves mixing, and thus the efficiency, of the sampling algorithms. The algorithms are tested on two diffusion processes: one with double-well potential drift and another with SINE drift. The new algorithm's accuracy and efficiency is compared with state-of-the-art hybrid Monte Carlo based path sampling. It is shown that in practical, finite sample applications the algorithm is accurate except in the presence of large observation errors and low to a multi-modal structure in the posterior distribution over paths. More importantly, the variational approximation assisted sampling algorithm outperforms hybrid Monte Carlo in terms of computational efficiency, except when the diffusion process is densely observed with small errors in which case both algorithms are equally efficient. © 2011 Springer-Verlag.

Diffuse beta-amyloid (A beta) deposits and neurons: in situ secretion or diffusion of A beta?

The association between diffuse-type beta -amyloid (AP) deposits and neuronal cell bodies in Alzheimer's disease (AD) and Down's syndrome (DS) could result from the secretion of AP from clusters of neurons in situ or the diffusion of A beta from cell processes, glial cells or blood vessels. To decide between these hypotheses, spatial pattern analysis was used to study the relationship between the degree of clustering of neuronal cell bodies and the presence of diffuse deposits in the temporal lobe of patients with DS. Significant clustering of neuronal cell bodies was present in 17/24 (71%) of brain areas studied. in addition, in 23/24 (96%) of brain areas, there was a positive correlation between the presence of diffuse deposits and the density of neurons. Hence, the data support the hypothesis that diffuse deposits develop in situ mainly as a result of the secretion of A beta by local clusters of neurons rather than by significant diffusion. Furthermore, the size of a diffuse deposit is likely to be determined by the number of neurons within a cluster which secrete A beta. The number and density of neurons could also be a factor determining the evolution of a diffuse into a mature amyloid deposit.

Show bizz becomes show buzz:how viral diffusion changes the traditional meaning-making process of a rising star

Studying the case of a young French rapper called Kamini, the authors show how the viral diffusion of a new creative product, such as a song, radically changes traditional meaning-making processes. Instead of the top-down approach in which product positioning is carefully constructed and transferred to consumers, marketers are faced with a bottom-up trend in which consumers increasingly participate in blogs and online forums to talk about products (thus, creating and diffusing meaning) before any marketing action is undertaken. Our study aims to understand the interactions and tensions between market forces that result from this pro-active role of the consumer.

ICT innovation diffusion in small logistics service providers:an empirical study

For companies competing in highly dynamic markets, innovation is considered a fundamental component of a successful business as it allows companies to sustain profit margins, sales growth and reduce competitors’ pressures. Information and communication technology (ICT) is essential innovation enablers especially in service companies. The focus of the paper is on the analysis of the role of ICT in innovation processes of small third-party logistics service providers (3PLs). On the basis of quantitative evidence emerging from a recent survey carried out on the Italian 3PL market, the paper analyses how ICT is used to support innovation and the factors the inhibit/facilitate the usage of ICT in such companies. Implications for supply chain innovation management are derived from the research and managerial perspectives.

Microstructural development in aluminide diffusion coatings on nickel-base superalloy single crystals

An investigation has been made of the microstructural stability of aluminide diffusion coatings during post-coating thermal exposure. This study has employed edge-on transmission electron microscopy to examine high-activity pack aluminised single crystals of a gamma prime strengthened nickel-base superalloy. The influence of exposure temperature, duration and atmosphere as well as the initial coating thickness has been assessed. Two major processes have been found to contribute to microstructural changes in the coating. These are, firstly, the transformation of the coating matrix (β-phase, nominally NiAl) to other Ni-Al based phases, especially γ' (nominally Ni3(Al, Ti)) and, secondly, the precipitation of chromium containing phases. The work has enabled the roles of three processes contributing to γ formation, namely: oxidation of the coating surface, interdiffusion with the substrate and ageing of the coating, to be understood. In addition, the factors leading to the formation of a sequence of chromium-containing phases have been identified.

Dynamic capabilities for CSR management:towards identifying common processes

Purpose – The objective of this paper is to address the question whether and how firms can follow a standard management process to cope with emerging corporate social responsibility (CSR) challenges? Both researchers and practitioners have paid increasing attention to the question because of the rapidly evolving CSR expectations of stakeholders and the limited diffusion of CSR standardization. The question was addressed by developing a theoretical framework to explain how dynamic capabilities can contribute to effective CSR management. Design/methodology/approach – Based on 64 world-leading companies’ contemporary CSR reports, we carried out a large-scale content analysis to identify and examine the common organizational processes involved in CSR management and the dynamic capabilities underpinning those management processes. Findings – Drawing on the dynamic capabilities perspective, we demonstrate how the deployment of three dynamic capabilities for CSR management, namely, scanning, sensing and reconfiguration capabilities can help firms to meet emerging CSR requirements by following a set of common management processes. The findings demonstrate that what is more important in CSR standardization is the identification and development of the underlying dynamic capabilities and the related organizational processes and routines, rather than the detailed operational activities. Originality/value - Our study is an early attempt to examine the fundamental organizational capabilities and processes involved in CSR management from the dynamic capabilities perspective. Our research findings contribute to CSR standardization literature by providing a new theoretical perspective to better understand the capabilities enabling common CSR management processes.