A new approach for long range corrections in molecular dynamics simulation with application to calculation of argon properties

In the application of long range corrections to calculation of potential energy and pressure utilizing molecular dynamics simulation, it is common to use a longer distance of cut-off to obtain more accurate results. The radial distribution function is assumed to be equal to unity in conventional terms for the long range corrections. The main objective of this work is to introduce a methodology for obtaining a correlation for radial distribution function from preliminary simulation runs and then use it in other simulation runs with a smaller ratio of cut-off distance to collision diameter. This, in turn, results in much lower CPU time requirements. The radial distribution function is considered as a convergent oscillating function. A number of simulations were performed in NVT ensemble with the ratio of cut-off distance to collision diameter set to five. Also, two sets of simulations with the ratio of two were performed. The first set used the conventional term of long range corrections and the second used the term presented in this work. With the ratio of two, calculating the pressure of argon in liquid and supercritical fluid states resulted in much lower CPU time and less error. In predicting pressure, applying the new correlation for radial distribution function reduced the CPU time by a factor of twelve. Assuming cut-off distance to be twice the collision diameter also resulted in a decrease of nearly 4% in relative error. In addition, the effect of this modification on argon properties: such as heat capacity, thermal pressure coefficient and isothermal compressibility are investigated using molecular dynamics simulation data in fluctuation formulas.