Multi-objective forcefield parameterization for thermal transport in 2D materials

Forcefields for the calculation of thermal properties of nanomaterials must be parameterized to match empirical material properties. Here, the third generation of the Non-Dominated Sorting Genetic Algorithm (NSGA-III) is used to construct forcefields to model 2D semiconducting materials by optimizing structure parameters (lattice constants), mechanical properties (elastic modulus) and vibrational behavior (phonon dispersion curve) from ab initio simulations. The algorithm is a parallelized, cross-platform workflow, written in C, that uses GULP as the engine for validating constructed forcefields. NSGA-III is a reference-point-based many-objective algorithm emphasizing population members that are non-dominated, yet close to a set of supplied reference points. It can handle up to fifteen variables and reference point tuning allows for user tuned diversity in the end product. This will allow for future expansion into forcefields finely-tuned over additional variables.