Brown University Center for Statistical Sciences Seminar
Bloomberg School of Public Health, Johns Hopkins University | |
4th Floor Conference Room |
Abstract: The form of the relationship between PM10 and mortality is critical for public health and regulatory decisions. We analyzed data on more than half of the U.S. deaths for the period 1987-1994, to determine how day-to-day variation in mortality counts varies with daily levels of PM10.
We developed hierarchical spline models. At the first stage, we estimated pollution-mortality dose-response curves by modeling the logarithm of the expected value of daily mortality as a function of air pollution using natural cubic splines with unknown number and locations of knots. At the second stage, we built spatial models to investigate the heterogeneity of dose-response curves across cities and across geographical regions.
Overall, we found that the pooled concentration-response relationship for the nation was linear, with little evidence for deviation from linearity down to the lowest levels.
These models have been developed to provide national-level estimates about the health effects of air pollution on which future regulatory decisions can be based more soundly.
Joint work with Scott L. Zeger, Michael Daniels, and Jonathan M. Samet.
*Reception following seminar at 167 Angell Street, 2nd Floor conference room.
Joint Seminar
Materials and Mechanical Engineering Divisions
Abstract: Selective electrochemical dissolution of silver from silver/gold alloys is a method to manufacture porous gold with ligament widths as small as 5 nm. This material has exciting potential in sensor and other xyz-on-a-chip applications that require ultra-high surface areas. Generally, selective dissolution involves a complex interplay between surface diffusion and chemical etching at the alloy/electrolyte interface, and the central issue of the formation mechanism of nanoporosity has been why doesn't gold simply uniformly accumulate on the surface as silver is etched away, blocking further dissolution? We will review our recent developments (experiments, lattice computer simulations, and continuum modeling) that show nanoporosity results from a non-equilibrium pattern forming instability operative on the alloy surface during dissolution. Pores form because the more noble atoms are chemically driven to aggregate into two-dimensional clusters via a spinodal decomposition process at the solid-electrolyte interface. Together, these processes evolve a characteristic length scale predicted by our continuum model, a surprisingly general result also related to many systems of crystal growth.
Brown Analysis Seminar
LATE ANNOUNCEMENT
Special PDE Seminar
(with P.A. Markowizh, G. Walansky) | |
**PLEASE NOTE SPECIAL DAY AND LOCATION FOR TODAY ONLY** |
Special Joint LCDS/Stochastic Systems Seminar
**PLEASE NOTE CHANGE OF DAY, TIME & PLACE FOR THIS WEEK ONLY** |
Abstract: We will discuss some recent results about stochastic averaging of Hamiltonian systems. We will argue that the appropriate spaces for looking at such problems are stratified spaces. We will understand various behaviors at the junction of these strata (i.e., glueing conditions) and connect some ideas of Khasminskii to the glueing conditions. The results will combine ideas from stochastic processes with dynamical systems.
Scientific Computing Seminar
Abstract: Dynamic materials are spatio-temporal composites formed from materials which are distributed on a microscale in space and in time. These materials are of particular interest when we want to affect the influences of long wave disturbances. A dynamic disturbance on a scale much greater than the scale of a spatio-temporal microstructure will perceive this formation as a new material with its own effective properties. With spatio-temporal variability in the material constituents, we shall be able to effectively control the dynamic processes by creating effects that are unachievable through purely spatial (static) material design. For example, by appropriately controlling the design factors of a dynamic composite, it is possible to selectively screen large domains in space-time from the invasion of long wave disturbances. One is also able to eliminate the existence of the cut-off frequency in electromagnetic waveguides.
In this talk, we focus on the direct numerical simulation of wave motion through dynamic composites. Work is presented for laminated materials where the material properties are periodic and move velocity V. When |V| is less than the individual characteristic speeds, a_i, of the individual materials, our results validate the effective wave speeds predicted by homogenization theory, and we are able to illustrate the screening effect. However, for the case when |V|>a_i, we show that there is inherent short wave instability. Long wave initial distrubances which are stable analytically are degraded in the computation due to the introduction of shorter wave modes coming from round-off and truncation errors. In the talk, we will give our resolution to this problem, present numerical results, and give applications of dynamic materials.
<--- 2001 Index