Molecular dynamics (MD) simulations are useful for a broad range of applications, including protein folding studies 1 , drug discovery 2 , and the determination of liquid structure and properties 3. Various approaches have been taken to improve the ability of simulations to explore the thermodynamically relevant parts of configuration space, including hardware advancements 4-9 and more effective sampling algorithms 10-14. While these efforts have dramatically improved our ability to generate well-converged results, errors persist in simulations, as highlighted for example, in the SAMPL series of blinded prediction exercises 15-21. Therefore, attention is now turning again to the potential functions, or force fields (FF), as sources of error, as recently reviewed 22 .

Force fields used in molecular simulations contain numerical parameters, such as Lennard–Jones (LJ) parameters, which are assigned to the atoms in a molecule based on a classification of their chemical environments. The number of classes, or types, should be no more than needed to maximize agreement with experiment, as parsimony avoids overfitting and simplifies parameter optimization. However, types have historically been crafted based largely on chemical intuition, so current force fields may contain more types than needed. In this study, we seek the minimum number of LJ parameter types needed to represent the key properties of organic liquids. We find that highly competitive force field accuracy is obtained with minimalist sets of LJ types; e.g., two H types and one type apiece for C, O, and N atoms. We also find that the fitness surface has multiple minima, which can lead to local trapping of the optimizer.

We utilize a previously described Minimal Basis Iterative Stockholder (MBIS) method to carry out an atoms-in-molecules partitioning of electron densities. Information from these atomic densities is then mapped to Lennard-Jones parameters using a set of mapping parameters much smaller than the typical number of atom types in a force field. This approach is advantageous in two ways: it eliminates atom types by allowing each atom to have unique Lennard-Jones parameters, and it greatly reduces the number of parameters to be optimized. We show that this approach yields results comparable to those obtained with the typed GAFF force field, even when trained on a relatively small amount of experimental data.