Brown University Center for Statistical Sciences Seminar
Abstract: Using computational methods to help understand gene regulation is one of the major challenges in the post-genome era. This talk describes some of our recent modeling and algorithmic efforts in the identification of regulatory binding motifs in noncoding DNA sequences. We mainly focus on the following topics: a) discovering correlated position pairs; b) modeling gene regulatory modules; and c) computational and modeling challenges. The new module-finding algorithm finds approximately 69\% of the possible experimentally reported transcription factor binding sites and 85\% of the possible modules in a test data set.
*Joint work with Chip Lawrence, Bill Thompson, Wyeth Wasserman, Xiaole Liu, and Mayetri Gupta **Sponsored by the C.V. Starr Foundation Lectureship Fund ***Co Sponsored by The Center for Genetics and Genomics
Brown Analysis Seminar
Scientific Computing Seminar
Abstract: We discuss the development, validation and implementation of a multidomain spectral method for direct numerical simulation (DNS) of turbulent compressible (carrier) flows laden with a dispersed phase of solid particles or liquid droplets. The DNS of two-phase, compressible flow has only received attention in simple geometries, such as homogeneous turbulence and a temporally developing mixing layer, due to computational restrictions. Recent enhancements in computational power and parallel programming have made the simulation in more complex geometries a possibility. However, the numerical schemes that are suitable for such simulations are somewhat recent.
The staggered-grid, multidomain, spectral method is employed because of its high accuracy, its ability to deal with complex geometries, and its local character which is favorable for parallel implementation. Specific numerical issues that were investigated are the extension of the code for simulation of three-dimensional flows, and the development of boundary condition treatments, a particle tracking algorithm and parallel algorithms. The performance of the method for the simulation of problems with a large range of scales is investigated through various 1-D, 2-D and 3-D simulations, including a running wave problem (advection equation), the `turbulent-like' transitional flow over a square rectangular cylinder, an isotropic decaying turbulence, the turbulent flow in a channel, and the the transitional backward-facing step flow with or without a countercurrent shear manipulation.
PDE Seminar
Abstract: Stieltjes' solution of the moment problem can be interpreted as the solution of an inverse scattering problem, and it provides a method of constructing the multi-peakon solutions of the Camassa-Holm equation, analogous to the multi-soliton solutions of the KdV equation. (This is joint work with Richard Beals and Jacek Szmigielski.)
Department of Mathematics Colloquium
<--- 2003 Index