Empirical Modeling of Nanoscale Dynamics using Solution Mapping:

 

In order to facilitate and expand the use of complex dynamic simulations for engineering tasks, solution mapping is a useful approach that is applied by many disciplines due to the need to reduce computations of complex systems. But there is no consistent method to determine how to divide the domain of the model or which regression functions to use and get a better accuracy with the model. My research is focused on the understanding of the errors in solution mapping using spatial stastics to model regression error, combining aspects of regression and interpolation of the sampled data points to make an unbiased optimal prediction in the solution map, and also computes the expected error in the fit. This approach to empirical modeling has recently gained interest for static models, under the name of kriging, but it has not been applied in a dynamics context, like nanoscale systems require.

Page updated on: November 8, 2006