In contrast to multiscale methods, which often encompass multiple simulation techniques, multiresolution models use the same general modeling technique with representations of a system at varying length and time scales. We present work on the reverse-mapping component of a complete framework for modeling of semidilute polymer solutions, encompassing both coarse-graining and reconstruction algorithms. The reconstruction process uses probability distributions obtained from the coarse-graining procedure with respect to bond lengths, angles, and torsions, as well as the non-bonded potentials, to reconstruct a more more detailed version of a coarse-grained polymer consistent both with geometric constraints as well as energetic considerations. Using a quality estimator model for the reconstruction process, we can successfully �reverse map� entire atomistic configurations from coarse-grained descriptions. Applications for the reconstruction of atomistic models from united-atom models and detailed models from coarse-grained models will be shown. In particular, we show how a 'learning-based' methodology for reconstructing the chains provides for a more consistent and gradual reduction in the energy of a reverse-mapped system, without the need for randomly finding extremely low energy states, as can occur when a non-learning based approach is used.  We demonstrate our approach on systems containing up to 106 monomers.

Library-based reverse mapping scheme for high-molecular-weight polymer chains

248th National Meeting (2014)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.