Noura Dawass

3.4. R ESULTS AND DISCUSSION 3 61 RDF correction methods. In the table, the differences (%) are computed using Eq. (3.12) . Using the RDF without a correction results in considerable differences in the values of G f , especially for the smallest simulation box with L box = 10. Out of the three methods, the van der Vegt correction [113] leads to the low- est differences in the KBIs results in the thermodynamic limit. In Figure 3.8 we show that the method of Cortes-Huerto et al. [83] is similar to the van der Vegt correction [113] when estimating the function g ∞ αβ ( r ). Still, when computing the KBIs the van der Vegt correction provides lower differences than the method of Cortes-Huerto et al. [83] , which assumes that the finite–size correction of the RDF is independent of r . Also, the van der Vegt correction is fairly simple to im- plement, and the corrections are applied to one simulation for each size, unlike the 1/ N correlation, where for each size two simulations are required. Another shortcoming of the latter method is its numerical inaccuracy when correcting the RDF, resulting in inconsistency with regard to predicting the KBIs. Further- more, we apply the correction method proposed by Cortes-Huerto et al. [83] to the same WCA mixture. The method enhances the computation of the integrals from finite systems, but the differences (%) in the KBI computations are not low- ered as in the case of the van der Vegt correction.