Mapping free energy landscapes with radial basis functions
Luca Maragliano (CIMS)

Free energy calculations are of central importance in Molecular Dynamics (MD) simulations. Most of the currently used techniques rely on regularly spaced grids and/or extensive sampling, and they become inadequate when the free energy we want to compute is in more than two dimensions. I will describe a new method to map free energy landscapes in several dimensions based on an variational interpolation technique using radial-basis functions. The method will be illustrated on simple examples and then used in a MD example to reconstruct the four dimensional free energy landscape of the solvated alanine dipeptide molecule.

The work is joint with Eric Vanden-Eijnden.