Molecular: Hierarchical Multiscale Modeling with a new class of force fields

We propose a novel methodology to produce a new class of force fields (FF) for predictive molecular simulations and computer modeling, based on large-scale ab initio computer simulations (with no empirical input), combined with mining from structural data bases such as CSD and PDB. The FFs are effective potentials, obtained in a by us developed inverse Newtonian scheme called the Inverse Monte Carlo (IMC). IMC solves completely and uniquely the inverse problem by reversing the radial distribution functions, obtained in ab initio (or other types of simulations) to pair potentials to be used as a FF in conventional classical simulations. The same procedure can be applied on crystal structures from data bases. A matrix of pair potentials is obtained as tables between all chosen pairs of interaction sites and the obtained FF is rather different from those used today. There is no a priori functional form, or difference between bonded and non-bonded interactions, and most importantly obtained at condensed state conditions at specified temperatures etc. We are currently developing our own tight-bindinglike DFT software to be used as an MD engine. We thereafter develop the same methodology even further and suggest a genuine hierarchical scheme for multi-scale modelling from nuclei and electrons to soft matter and beyond by introducing successive coarse-graining and solvent and/or environment mediated potentials.