Nonparametric Bayesian sparse factor models with application to gene expression modeling

A nonparametric Bayesian extension of Factor Analysis (FA) is proposed where observed data $\mathbf{Y}$ is modeled as a linear superposition, $\mathbf{G}$, of a potentially infinite number of hidden factors, $\mathbf{X}$. The Indian Buffet Process (IBP) is used as a prior on $\mathbf{G}$ to incorporate sparsity and to allow the number of latent features to be inferred. The model's utility for modeling gene expression data is investigated using randomly generated data sets based on a known sparse connectivity matrix for E. Coli, and on three biological data sets of increasing complexity.

Unsteady transition in an axial-flow turbine. Part 2. Cascade measurements and modeling

Part 1 of this paper reanalyzed previously published measurements from the rotor of a low-speed, single-stage, axial-flow turbine, which highlighted the unsteady nature of the suction surface transition process. Part 2 investigates the significance of the wake jet and the unsteady frequency parameter. Supporting experiments carried out in a linear cascade with varying inlet turbulence are described, together with a simple unsteady transition model explaining the features of seen in the turbine.

Modeling of unsteady transitional boundary layers

In turbomachinery, a considerable proportion of the blade surface area can be covered by transitional boundary layers. This means that accurate prediction of the profile loss and boundary layer behavior in general depends on the accurate modeling of the transitional boundary layers, especially at low Reynolds numbers. This paper presents a model for determining the intermittency resulting from the unsteady transition caused by the passage of wakes over a blade surface. The model is founded on work by Emmons (1951) who showed that the intermittency could be calculated from a knowledge of the behavior of randomly formed turbulent spots. The model is used to calculate the development of the boundary layer on the rotor of a low Reynolds number single-stage turbine. The predictions are compared with experimental results obtained using surface-mounted hot-film anemometers and hot-wire traverses of the rotor midspan boundary layer at two different rotor-stator gaps. The validity and limitations of the model are discussed.

Intermittency transport modeling of separated flow transition

An intermittency transport model is proposed for modeling separated-flow transition. The model is based on earlier work on prediction of attached flow bypass transition and is applied for the first time to model transition in a separation bubble at various degrees of free-stream turbulence. The model has been developed so that it takes into account the entrainment of the surrounding fluid. Experimental investigations suggest that it is this phenomena which ultimately determines the extent of the separation bubble. Transition onset is determined via a boundary layer correlation based on momentum thickness at the point of separation. The intermittent flow characteristic of the transition process is modeled via an intermittency transport equation. This accounts for both normal and streamwise variation of intermittency and hence models the entrainment of surrounding flow in a more accurate manner than alternative prescribed intermittency models. The model has been validated against the well established T3L semicircular leading edge flat plate test case for three different degrees of free-stream turbulence characteristic of turbomachinery blade applications.

Modeling of MOS-side carrier injection in trench-gate IGBTs

A compact trench-gate IGBT model that captures MOS-side carrier injection is developed. The model retains the simplicity of a one-dimensional solution to the ambipolar diffusion equation, but at the same time captures MOS-side carrier injection and its effects on steady-state carrier distribution in the drift region and on switching waveforms. © 2007 IEEE.

Gaussian mixture modeling with Gaussian process latent variable models

Density modeling is notoriously difficult for high dimensional data. One approach to the problem is to search for a lower dimensional manifold which captures the main characteristics of the data. Recently, the Gaussian Process Latent Variable Model (GPLVM) has successfully been used to find low dimensional manifolds in a variety of complex data. The GPLVM consists of a set of points in a low dimensional latent space, and a stochastic map to the observed space. We show how it can be interpreted as a density model in the observed space. However, the GPLVM is not trained as a density model and therefore yields bad density estimates. We propose a new training strategy and obtain improved generalisation performance and better density estimates in comparative evaluations on several benchmark data sets. © 2010 Springer-Verlag.