The pair- and triplet-correlation functions g12[≡ g(r12)] and g123[≡ g(r12, r13, r23)] are used to make a superposition approximation g123g124g134g234 / (g12g13g14g23g24g34) to the quadruplet correlation function g1234. Introduction of this approximation into the second equation of the Born–Green–Yvon (BGY) hierarchy makes it possible to truncate the hierarchy. The resulting equations are a pair of simultaneous integro-differential equations involving g12 and g123, which hereafter are referred to as the BGY2 equations. The BGY2 theory was further elucidated by the use of the hard-sphere potential to solve the corresponding BGY2 equations at two selected values (10 and 15) of λ, which is used as the independent variable and is equal to 4πρg(σ) (ρ = density, σ = sphere diameter). The densities {ρ = λ / [4πg(σ)]} corresponding to these values of λ can be established from the BGY2 solutions of g(r) at contact distance of two spheres. They are found to be, respectively, 0.298ρ0 and 0.372ρ0 (ρ0 = the close-packed density). At these densities, a separate program, which uses a Monte Carlo (MC) method, was used to evaluate the radial distribution functions for 500 hard spheres. The MC g(r) can be regarded as exact within the observed standard deviations; hence, adequacy of various approximate theories of g(r) can be tested against it. This was done here, not only for the BGY2 g(r) but also for the BGY1 g(r) obtained after truncating the BGY hierarchy by using the Kirkwood superposition approximation for g123 as well as for the Percus–Yevick (PY) g(r). At ρ = 0.298ρ0, the comparison gave values of the g(r) which are vastly improved over values calculated from either the BGY1 g(r) or the PY g(r) and which agree within each other's numerical uncertainties with the MC g(r) at most of the interparticle distances. Nearly perfect agreement (again within numerical uncertainties of the two) of the BGY2 g(r) and the MC g(r) at ρ = 0.372ρ0 was also observed for r < 1.4σ. For r > 1.4σ, the BGY2 g(r) deviates slightly (i.e., 1.6% or less) from the MC g(r) which appears to follow slightly closer to the PY g(r). The discrepancies between the BGY2 and the MC g(r) may be attributable to a large extent to the iterative scheme used in solving BGY2 equations as well as use of the inaccurate density in obtaining the MC g(r). Over-all improvement of the BGY2 g(r) over the BGY1 g(r) is striking. In particular, the pressures calculated from the BGY2 g(r) by the use of the virial theorem agree within statistical uncertainties with the MC pressures at the densities given above.
Read full abstract