Fast Energy Calculations and Large-Scale Inverse Problems in Imaging

Abstract: 

Fast Energy Calculations and Large-Scale Inverse Problems in Imaging

This researcher is developing fast, accurate, methods for computing energies of use in Monte Carlo simulations of molecular dynamics. These methods involve high-precision approximation of pairwise interactions between thousands to millions of atoms. The code is being developed in FORTRAN. The researcher is using MSI for possible parallelizations of the code and for access to the Intel FORTRAN compiler, which is not available at his home institution.

A second project in remote sensing involves computational inverse problems in imaging. The optimization problems and related PDEs can be large enough to require parallelization for computational tractability. A decomposition of the problem over a convenient set of orthogonal basis functions introduces embarrassing parallelizability. This code is currently being developed in MATLAB.

Return to this PI's main page.

 

Group name: 
hoftt