Coarse graining of the fully atomic methane models to monatomic isotropic models using relative entropy minimization

This work presents the relative entropy minimization method for the construction of a series of coarse grained (CG) methane models based on the four popular methane atomistic models (RISM, LJ (12-6), OPLS and MOPLS). Each CG methane molecule is represented by one bead that interacts through short-ranged isotropic interactions. The calculated dynamic and structural properties of CG model show reasonable agreement with the results of the corresponding atomistic model or experiment. To evaluate the predictive power of the CG model for prediction of mass transfer and the structural properties, the relative error of diffusion coefficient RED and the residuals of radial distribution function RESRDF are introduced. Numerical results demonstrate that the CG models can quantitatively reproduce the structural and dynamic properties of the methane fluid. Furthermore, the CGMOPLS could reproduce quite well the radial distribution function, velocity autocorrelation function and diffusion coefficient. Interestingly, we find that the CG potential not only depends on the reference thermodynamic state conditions but also relies on the corresponding atomistic model.