Abstract
A kernel-based method (kernelized minimal distributed charge model - kMDCM) to represent the molecular electrostatic potential (ESP) in terms of off-center point charges whose positions adapts to the molecular geometry. Using Gaussian kernels and atom-atom distances as the features, the ESP for water and methanol is shown to improve by at least a factor of two compared with point charge models fit to an ensemble of structures. Combining kMDCM for the electrostatics and reproducing kernels for the bonded terms allows energy-conserving simulation of 2000 water molecules with periodic boundary conditions on the nanosecond time scale.