(* Copyright (C) 2017 M.A.L. Marques 2019 Susi Lehtola This Source Code Form is subject to the terms of the Mozilla Public License, v. 2.0. If a copy of the MPL was not distributed with this file, You can obtain one at http://mozilla.org/MPL/2.0/. *) (* type: mgga_exc *) $include "mgga_xc_b98.mpl" (* Fitted parameters, given below Eq. 50 *) edmgga_mu := 0.21: edmgga_a := 0.704: (* Eq. 40: F_x'(0) *) edmgga_fp0 := 27*edmgga_mu/50: (* Eq. 27: lim_(x -> -infinity) F_x(x) *) edmgga_fminf := 1/3 * (4*Pi^2/3)^(1/3): (* Eq. 44 *) edmgga_b := 2*Pi/9 * sqrt(6/5): (* Eq. 47 *) edmgga_c1 := edmgga_fminf: (* Eq. 48 *) edmgga_c2 := 1 - edmgga_c1: (* Eq. 49 *) edmgga_c3 := edmgga_c2*sqrt(2*edmgga_a)/(2*edmgga_b): (* Eq. 50. There's some problem in the equation, since the first line, 1/edmgga_c3^3 * (edmgga_c3^2 - edmgga_fp0/(2*edmgga_b^2)) * edmgga_c2, does not agree with the second line. The first line evaluates to a negative value, which makes the functional blow up. The second line yields values that are close to the reported ones, but Tao might have used a more accurate numerical value in his implementation since the agreement is not perfect. Unfortunately there's no way to check this as he has passed away. *) edmgga_c4 := 1/edmgga_c3^3 * (edmgga_c3^2 - 0.09834*edmgga_mu): (* Eq. 46 *) (* Apparently, edmgga_x can sometimes be smaller than 0, as Qb can be very negative. The expansion for -infinity has an error of 1e-13 *) edmgga_x_large := y -> -1/(2*y) + 1/(8*y^3): edmgga_x := Qb -> my_piecewise3( Qb < -1e4/edmgga_a, edmgga_x_large(m_min(edmgga_a*Qb, -1e-4/edmgga_a)), edmgga_a*Qb + sqrt(1 + (edmgga_a*Qb)^2) ): (* Eq. 45 *) edmgga_f_x := x -> edmgga_c1 + edmgga_c2*x/(1 + edmgga_c3*sqrt(x)*arcsinh(edmgga_c4*(x - 1))): (* the evalf is absolutely necessary to avoid exceptions *) edmgga_f := (x, u, t) -> edmgga_f_x(edmgga_x(b98_q(x, u, t))): f := (rs, z, xt, xs0, xs1, u0, u1, t0, t1) -> mgga_exchange(edmgga_f, rs, z, xs0, xs1, u0, u1, t0, t1):