Economic simplex optimization for broad range property prediction: Strengths and weaknesses of an automated approach for tailoring of parameters

The need for predicting physical compound properties supplementing experimental data is considerable. Nowadays a wide range of classical simulation techniques is available for computing a multitude of such properties with acceptable effort. We here give a field report about our approach, which was to fit an initial model to a single point in the phase diagram. By way of accessing commonly available experimental values we developed a compound-specific force field via simplex optimization. For predicting the desired properties of the novel model we did engage classical equilibrium as well as reverse non-equilibrium molecular dynamics in combination with Monte Carlo methods and report here the performance of these methods in detail for the example compound ethylene oxide. We find that the new model describes the experimentally observed behavior of the test compound ethylene oxide (EO) very well in the molecular dynamics section. However, when applying the simplex optimized model to the Monte Carlo section, the limits of transferability become apparent.