Brown University -
Joint Materials/Solid Mechanics Seminar Series
Abstract: The borderlands between elasticity and hydrodynamics lead naturally to a number of moving boundary problems in elastohydrodynamics. I will discuss some phenomena in this rich area involving extreme geometries: the flutter of a slender flag in a breeze (and its relation to fish swimming), the lift on a soft fluid-lubricated solid sliding/rolling near a wall (and its relation to joint lubrication), and the dynamics of fluid-filled tissues (and its relation to rapid movements in some plants).
Cognitive & Linguistic Sciences
Spring 2005 Colloquium and CG233 Speakers
Stanford University | |
Brown University Center for Statistical Sciences Seminar
Abstract: The analysis of the high-dimensional data generated by DNA microarrays poses challenge to standard statistical methods.
In this talk I will describe how Bayesian methodologies for variable selection can be successfully employed in the analysis of genomics data. Stochastic search techniques for variable selection, originally developed for regression settings, have been extended to a variety of different models, including recent new developments for classification and clustering problems. I will describe the key ideas of these statistical methods and present two applications in genomics.
In the first application microarray data are employed to refine the search for regulatory motifs. A linear regression model is used to relate microarrays to sequence data. Stochastic search techniques allow the identification of putative binding sites for regulatory factors. The second application focuses on clustering problems involving microarray data where the interest is in the identification of distinct subtypes of a disease together with the detection of the discriminating genes.
Molecular classes defined on a small number of genes can lead to a better understanding of the underlying biological processes. In addition, the selected genes can serve as biomarkers for improved diagnosis and targets for therapeutic intervention. I will formulate this clustering problem in terms of a multivariate normal mixture model with an unknown number of components and use reversible jump MCMC to define a sampler that moves between different dimensional spaces. The flexibiliby of the prior models allows to incorporate biological information.
Center for Fluid Mechanics and The Fluids, Thermal and Chemical Processes Group of The Division of Engineering Seminar Series
Abstract: Wave breaking on the surface of the ocean results in significant transfer of mass, momentum, heat and energy across the air-sea interface. However, understanding the physics of these phenomena remains a challenge, as both the measurement and simulation of the relevant processes are highly complex. Recent numerical studies have begun to investigate the air-water coupling during the breaking process using level set methods (Hendrickson, 2004), but relatively few experimental studies have looked at flow on both sides of this interface. In order to more clearly understand the whole picture and to generate accurate models, experiments and simulations which consider the physics on both the air and water-side of the air-sea interface are necessary.
This talk presents experiments on small-scale, steep breaking waves forced by shoaling the waves up a ten degree slope to a level plateau. Breaking waves are generated by a computer controlled, paddle-type wave maker at one end of small (250cm long, 15.25 cm wide, 20.3 cm deep) acrylic tank. A beach is place at the far end of the tank to absorb reflections. The waves generated in this tank were smaller in scale than typical wave breaking experiments, in order to compare with numerical simulations by Hendrickson which were on
the order of Reynolds number $10^{3}-10^{4}$. Reynolds number in this case is defined as $Re = \frac {\rho \sqrt g \lambda^{3}}{\mu} $ ,
where $ \rho $ is the fluid density, $ \mu $ is the fluid dynamic viscosity, $ g $ is the gravitational constant, and $ \lambda $ is the characteristic wavelength of the breaking wave. Reynolds numbers for the experiments presented here are between $ 9 \times 10^{4}$ and $ 2 \times 10^{6}$.
Qualitative and quantitative measurements were obtained through high-speed imaging techniques, including high speed PIV. While PIV seeding in the water was quite straight-forward, seeding in the air was quite complex. Air-seeding techniques such as atomized oil droplets are not appropriate in this study as the oil would negatively influence the surface tension significantly. Thus a water-soluble fog was used to seed the air, again no change in surface tension was recorded by addition of the fog. High speed video results show the formation of the vortex aft of the breaking wave and reveal strong counterclockwise vorticity in the air side of the interface. Conversely in the water, the majority of the vorticity has clockwise rotation for waves traveling from left to right. Figure 1 shows a snapshot of velocity (top) and vorticity (bottom) for a plunging breaker after the jet from the face of the wave has impacted the free surface and the air cavity has collapsed into smaller bubbles. Vector colors and length in the top plot represent magnitude of velocity from high (red) to low (blue). In the vorticity plot (below) the red represents clockwise vorticity and the blue counterclockwise vorticity. The red line shows a profile of the surface above which there is no water. Below this line is water except in a region of two-phase flow where air has been entrained due to the plunging jet.
This talk will present data from the experiments and discuss the experimental issues relating to measurements on the air-side of the air-sea interface. Challenges remain to refine air-side visualization method with improved seeding method and the implementation of a two-laser/two-camera setup with high-speed imaging capabilities. Data from wave probes and PIV measurements can then be used to quantify energy dissipation and transfer across air-water interface and the turbulent structures in the flow.
References : K.L. Hendrickson, "Navier-Stokes Simulations of Steep Breaking Water Waves with a Coupled Air-Water Interface," PhD Thesis, Massachusetts Institute of Technology, 2004.
Brown University Division of Biology and Medicine -
Center for Statistical Sciences Seminar Series
Candidate for Open Rank Professor (research track) in the Public Health/Department of Community Health | |
Abstract: We consider semiparametric regression for high dimensional data, e.g., microarrays. This model relates a normal clinical outcome to clinical covariates and gene expressions, where the clinical covariate effects are modelled parametrically and gene expression effects are modelled nonparametrically using kernel machines. The kernel machine estimate of the nonparametric function allows for the possibility that the number of genes might be large, each genetic effect may be non-linear, and many genes are likely to interact with each other. The regression coefficients and nonparametric function are estimated through the penalized maximum likelihood method (PML). Using results from constrained optimization theory we show that the PML estimation is the same as that from a linear mixed model: both the regression coefficients of the parametric effects and the kernal machine estimator of the nonparametric function can be obtained using the best Linear Unbiased Predictor (BLUP) in linear mixed models. The smoothing parameter and the kernal scale parameter can be estimated as variance components using REML in linear mixed models. A bootstrap test is developed to test for the nonparametric function effect. The methods are illustrated using a prostate cancer data set and evaluated using simulation studies.
Center for Computational Molecular Biology Seminar
Currently Visiting at the Bauer Center for Genomics Research and The Division of Engineering and Applied Science at Harvard University | |
Abstract: High-throughput genome-wide molecular assays have become central to molecular biology. These assays probe cellular networks from different perspectives and provide rich and diverse data, posing the challenge of developing methodologies for extracting meaningful biological insights. The challenge for computational biology is to provide methodologies for transforming high-throughput heterogeneous datasets into biological insights about the underlying mechanisms. Integration of data from assays that examine cellular systems from different viewpoints can lead to a more coherent reconstruction, reduce the effects of noise, and provide new knowledge about the relevant biological entities and processes.
One class of approaches to answer this challenge builds on probabilistic graphical models. Such models provide a concise representation of complex cellular networks models by composing simpler sub-models. Procedures based on well understood principles for inferring such models from data facilitate a model--based methodology for analysis and discovery. In this talk I will attempt to discuss few recent projects that use the language of probabilistic graphical models to model and understand gene regulation and function from genomics datasets.
Time permitting, I will cover the following three projects:
* Reconstructing "realistic" models of transcriptional regulation circuits (joint work with Iftach Nachman and Aviv Regev).
* Modeling transcription factor binding sites without assuming independencies between binding site positions (joint work with Yoseph Barash, Tommy Kaplan, and Gal Elidan)
* Learning the structure of protein-protein interaction networks (joint work with Ariel Jaimovich and Gal Elidan)
Brown University Mathematics Department - Distinguished Lecture Series
Scientific Computing Seminar
Abstract: An original mixed finite-difference/spectral method based upon a Fokker-Planck equation is proposed for the numerical simulation of non-homogeneous flows of suspensions of finitely extensible nonlinear elastic (FENE) dumbbells. Since the configuration distribution distribution function $\psi$ for the dumbbells varies as a function of time and in both physical and configuration space careful attention has been paid to the discretization scheme for derivatives of $\psi$ in these variables, the domains of definition of the physical and configuration coordinates beine inter-dependent. Numerical results for start-up planer Poiseuille flow are in excellent quantitative agreement with those of a stochastic simulation and, for comparable levels of accuracy, are much less CPU expensive. Theoretical results for the polymeric stress fields under equilibrium conditions are verified numerically and generalize earlier results for Hookean dumbbells by Brunn and Grisafi [P. O. Brunn and S. Grisafi, Chem. Eng. Commun., 36 (1985) 367-383].
Some interesting parallels with micro-channel blood flow will be highlighted.
Departement de mathematiques et de statistique Universite de Montreal CP 6128 succ. Centre-Ville Montreal QC H3C Canada
PDE Seminar
Brown University Department of Mathematics - Distinguished Lecture Series
Brown University Department of Mathematics - Distinguished Lecture Series
Brown University Division of Biology and Medicine - Center for Statistical Sciences Seminar Series
Candidate for Open Rank Professor (research track) in the Public Health Program/Department of Community Health | |
Abstract: This development of cancer is a complex process. In the last two decades, major genes have been discovered which play important roles in cancer development. Among the most notable is the discovery of the BRCA1 and BRCA2 genes for breast and ovarian cancer. I will talk about the investigation of cancer risks attributable to known major genes (``penetrance"). This includes an overview of the design and analysis of studies on rare disease genes, with an emphasis on studies based on high-risk family, that is families with a high incidence and/or early onset of the cancers of interest. I will specifically talk about the analysis of these studies using the ``retrospective approach" and the ``ascertainment conditional approach", the comparison of unbiasedness and efficiency, in the context of breast and ovarian cancer arising from deleterious mutations of the BRCA1 and BRCA2 genes.
<--- 2005 Index