|
Leland Jameson  Parallel computing offers the greatest possibility for growth in obtaining the most computations per second in scientific computations. Furthermore, very few interesting or challenging physical problemscontain physical information which is uniform in scale. Certainly this includes computations of ocean currents, air-sea interaction, etc. In order to obtain the best computational result on a given generation of computers it is necessary to introduce numerical schemes which dynamically adaptive. Given the large variety of scales present, wavelets provide a very natural mechanism to guide this adaptivity. However, adaptive multiscale numerical schemes on parallel computers is a new and very challenging area of science. I am working on such implementations in order to obtain the fastest numerical results on parallel computers. Information obtained through observations must be properly analyzed in order to obtain the maximum amount of physical information present in the data. This observational information might come from buoy data, satellite data or other sources, and the quantity of data can be very large. There are many classical methods for analyzing such data, but wavelets offer a new approach. Wavelets can analyze data in ways which were not possible just a few years ago. I am considering various wavelet applications to analyze observational data. Estimating subsurface ocean activity using only data obtained from satellites is a very challenging scientific problem. I am beginning an investigation into this area. |
